Università di Venezia

Transcription

Università di Venezia
Corso di Laurea
in Ingegneria Gestionale
Tesi di Laurea di Primo Livello
Analisi dello stato dell’arte
dei sistemi automatici e
flessibili di produzione
Relatore
prof. Maurizio Faccio
Laureando
Lucio Barbazza
Matricola 624626
Anno Accademico
2010-2011
S OMMARIO
La tesi si propone di approfondire le tematiche attinenti ai sistemi flessibili di
assemblaggio come componenti chiave di una produzione altamente efficiente.
A causa di lotti di produzione di dimensioni molto piccole e alla necessità di
prodotti personalizzati, è sempre più richiesta maggiore flessibilità e adattabilità
nelle mansioni.
Il lavoro risulta così strutturato:
- nel Capitolo 1 forniamo una breve descrizione dei sistemi logistici integrati e flessibili, che siano in grado di realizzare l’integrazione dei flussi
fisici e dei flussi informativi per garantire: sia un elevato livello qualitativo dei prodotti e del servizio ai clienti, tramite la riduzione del tempo
di risposta all’ordine e il contenimento dei costi di produzione; sia di far
fronte con adeguata flessibilità al cambiamento continuo della gamma produttiva, conseguente alla variabilità e alla personalizzazione dei gusti del
consumatore;
- nel Capitolo 2 raccogliamo una serie di articoli sullo stato dell’arte sui sistemi di assemblaggio e sulle tecniche di alimentazione flessibile degli ultimi
dieci anni. Un ostacolo fondamentale per l’applicazione economica dell’automazione flessibile è l’alimentazione delle parti componenti. Tradizionalmente tale compito viene demandato ai vibroalimentatori i quali hanno un
costo elevato e sono dedicati a un singolo componente. Di conseguenza
il numero di alimentatori dedicati richiesto nel caso di una elevata varietà di prodotti è pari al numero delle distinte parti che compongono il mix
produttivo. Ciò comporta un costo elevato degli investimenti giustificato
solo nel caso della produzione di massa. Inoltre metodi di alimentazione
dedicati risultano poco flessibili nel caso si utilizzino sistemi di montaggio
robotizzati dove si produce una vasta gamma di componenti in piccoli lotti;
- nel Capitolo 3 presentiamo in modo approfondito i lavori di M. Morioka, S.
Sakakibara (2010) e J. Krüger, T.K. Lien, A. Verl (2009);
- nel Capitolo 4 concludiamo la tesi esponendo in primo luogo i risultati
ottenuti. Successivamente vengono proposti alcuni suggerimenti per i lavori futuri, consapevoli che molto lavoro rimane da fare in merito ai sistei
ii
mi di produzione ed assemblaggio flessibile. Infine vengono tratte alcune
conclusioni.
I NDICE
Elenco delle tabelle
vii
Elenco delle figure
ix
1 Logistica integrata e flessibile
1.1 Definizione della funzione logistica . . . . . . . . . . . . . . . . . . .
1.2 Flessibilità come risposta alla variabilità . . . . . . . . . . . . . . . .
1.3 Analisi P-Q e tipologie di layout . . . . . . . . . . . . . . . . . . . . .
1.4 Group technology e i sistemi flessibili di fabbricazione . . . . . . . .
1.4.1 La group technology . . . . . . . . . . . . . . . . . . . . . . .
1.4.2 FMS - Flexible Manufacturing System . . . . . . . . . . . . .
1.4.3 Elementi costitutivi di un sistema FMS . . . . . . . . . . . . .
1.4.3.1 Macchine operatrici . . . . . . . . . . . . . . . . . .
1.4.3.2 Sistemi di movimentazione dei pezzi . . . . . . . .
1.4.3.3 Sistemi per la gestione degli utensili . . . . . . . . .
1.4.3.4 Logiche di controllo . . . . . . . . . . . . . . . . . .
1.4.4 Progettazione di un sistema FMS . . . . . . . . . . . . . . . .
1.5 Mezzi logistici nell’assemblaggio automatico . . . . . . . . . . . . .
1.5.1 Complessità del prodotto - struttura modulare . . . . . . . .
1.5.2 Unità di base . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.5.3 Numero di componenti . . . . . . . . . . . . . . . . . . . . .
1.5.4 Architetture di una linea di assemblaggio . . . . . . . . . . .
1.5.5 Elementi costitutivi di una linea flessibile di assemblaggio .
1.5.5.1 Attrezzi portanti del pezzo . . . . . . . . . . . . . .
1.5.5.2 Sistemi di trasporto degli attrezzi portanti . . . . .
1.5.5.3 Sistemi di alimentazione dei componenti . . . . . .
1.5.5.4 Stazioni di lavoro . . . . . . . . . . . . . . . . . . . .
1.5.5.5 Stazioni di controllo . . . . . . . . . . . . . . . . . .
1.5.6 Progettazione di una linea flessibile di assemblaggio automatico . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.5.6.1 Razionalizzazione del progetto dei prodotti secondo il Design For Assembling - DFA . . . . . . . . .
1.5.6.2 Razionalizzazione del processo produttivo . . . . .
1.5.6.3 Definizione del prodotto caratteristico . . . . . . . .
1
1
2
4
5
5
8
9
9
10
12
13
14
17
17
19
20
20
22
22
23
25
30
31
iii
32
32
32
33
iv
INDICE
1.5.6.4
1.5.6.5
1.5.6.6
1.5.6.7
1.5.6.8
Determinazione dei tempi e dei metodi di montaggio . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Scelta della cadenza della linea e dimensionamento delle stazioni . . . . . . . . . . . . . . . . . . . . .
Scelta del sistema di trasporto e di alimentazione
dei componenti . . . . . . . . . . . . . . . . . . . . .
Verifica della flessibilità e simulazione . . . . . . . .
Valutazione economica dell’investimento . . . . . .
33
33
34
34
34
2 Flexible automation state of the art
35
2.1 Flexible systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.1.1 A new cell production assembly system with human-robot
cooperation - 2010 . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.1.2 Bicriteria robotic operation allocation in a flexible manufacturing cell - 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.1.3 An analysis of cyclic scheduling problems in robot centered
cells - 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.1.4 Pure cycles in flexible robotic cells - 2009 . . . . . . . . . . . . 38
2.1.5 Cooperation of human and machines in assembly lines - 2009 38
2.1.6 A market approach to decentralized control of a manufacturing cell - 2009 . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.1.7 Modeling of machine failures in a flexible manufacturing
cell with two machines served by a robot - 2008 . . . . . . . 40
2.1.8 Effects of maintenance policies on the productivity of flexible manufacturing cells - 2006 . . . . . . . . . . . . . . . . . . 40
2.1.9 Design configuration for a mixed-model assembly system
in case of low product demand - 2006 . . . . . . . . . . . . . 41
2.1.10 Dynamic scheduling in flexible assembly system based on
timed Petri nets model - 2005 . . . . . . . . . . . . . . . . . . 41
2.1.11 A stochastic model for the analysis of a two-machine flexible manufacturing cell - 2005 . . . . . . . . . . . . . . . . . . 42
2.1.12 Image based 3D Surveillance for flexible Man-Robot-Cooperation
- 2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.1.13 A heuristic approach to batching and scheduling a single
machine to minimize setup costs - 2004 . . . . . . . . . . . . 43
2.1.14 Framework for designing a flexible cellular assembly system - 2004 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.1.15 Mini assembly cell for the assembly of mini-sized planetary
gearheads - 2004 . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.1.16 An approach to the design of a manufacturing cell under
economic considerations - 2002 . . . . . . . . . . . . . . . . . 45
2.1.17 Knowledge intensive Petri net framework for concurrent
intelligent design of automatic assembly systems - 2001 . . . 46
2.2 Techniques for flexible feeding . . . . . . . . . . . . . . . . . . . . . 47
2.2.1 Development and testing of a brush feeder - 2010 . . . . . . 47
INDICE
v
2.2.2
2.2.3
2.2.4
2.2.5
2.2.6
2.2.7
2.2.8
2.2.9
2.2.10
2.2.11
2.2.12
2.2.13
2.2.14
2.2.15
2.2.16
2.2.17
2.2.18
2.2.19
2.2.20
2.2.21
2.2.22
Design of a modular feeder for optimal operating performance - 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Approximation to the dynamics of transported parts in a
vibratory bowl feeder - 2009 . . . . . . . . . . . . . . . . . . .
On-line dimensional measurement of small components on
the eyeglasses assembly line - 2008 . . . . . . . . . . . . . . .
Dynamic modeling and experimental verification of a piezoelectric part feeder in a structure with parallel bimorph beams
- 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Advanced vision guided robotics provide “future-proof”
flexible automation - 2006 . . . . . . . . . . . . . . . . . . . .
Economic application of piezoelectric actuator in linear vibratory feeding - 2006 . . . . . . . . . . . . . . . . . . . . . .
A new type of parts feeder driven by bimorph piezo actuator - 2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A modular contactless feeder for microparts - 2005 . . . . . .
Piezo actuated vibratory feeding with vibration control - 2005
Development of a flexible and programmable parts feeding
system - 2005 . . . . . . . . . . . . . . . . . . . . . . . . . . .
Modal analysis and control of a bowl parts feeder activated
by piezoceramic actuators - 2004 . . . . . . . . . . . . . . . .
Simulation software for parts feeding in a vibratory bowl
feeder - 2003 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using air streams for part feeding systems - innovative and
reliable solutions for orientation and transport - 2003 . . . .
Mechatronic design of a flexible vibratory feeding system 2003 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chaotic dynamics of repeated impacts in vibratory bowl
feeders - 2002 . . . . . . . . . . . . . . . . . . . . . . . . . . .
Generic flexible assembly system design - 2002 . . . . . . . .
Orienting polyhedral parts by pushing - 2002 . . . . . . . . .
Development of a model for part reorientation in vibratory
bowl feeders with active air jet tooling - 2001 . . . . . . . . .
Flexible parts feeding for flexible assembly - 2001 . . . . . .
New Feeding System for High Speed Assembly of Small
Parts - 2000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Designing a parts feeding system for maximum flexibility 1997 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3 The article selected
3.1 Cooperation of human and machines in assembly lines . . . . . . .
3.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1.2 Human-machine cooperation in assembly lines - state of the
art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1.2.1 Hybrid assembly . . . . . . . . . . . . . . . . . . . .
3.1.2.2 Workplace sharing systems . . . . . . . . . . . . . .
47
47
48
49
50
50
51
51
51
52
53
53
54
54
55
55
55
56
56
57
57
59
59
59
60
60
62
vi
INDICE
3.2
3.1.2.3 Workplace and time sharing systems . . . . . . . .
3.1.3 Technologies for cooperative assembly . . . . . . . . . . . . .
3.1.3.1 Interfaces between human and machines . . . . . .
3.1.3.2 Robotics . . . . . . . . . . . . . . . . . . . . . . . . .
3.1.3.3 Sensors and actuators . . . . . . . . . . . . . . . . .
3.1.3.4 Safety systems . . . . . . . . . . . . . . . . . . . . .
3.1.4 Component supply and logistics . . . . . . . . . . . . . . . .
3.1.4.1 The materials supply challenge . . . . . . . . . . . .
3.1.4.2 Flexible gripping devices . . . . . . . . . . . . . . .
3.1.4.3 Intelligent gripping . . . . . . . . . . . . . . . . . .
3.1.4.4 General feeding subtasks . . . . . . . . . . . . . . .
3.1.4.5 Design for assembly applied to feeding problems .
3.1.4.6 Sub-problems of the feeding task . . . . . . . . . . .
3.1.4.7 Feeders for small parts in large volume production
3.1.4.8 Large part feeding . . . . . . . . . . . . . . . . . . .
3.1.4.9 Orientation mechanisms . . . . . . . . . . . . . . . .
3.1.4.10 Flexible component feeders . . . . . . . . . . . . . .
3.1.4.11 Flexible automated logistics . . . . . . . . . . . . . .
3.1.4.12 Intelligent warehousing . . . . . . . . . . . . . . . .
3.1.4.13 Parts feeding with intelligent assist devices . . . . .
3.1.5 Organizational and economic aspects . . . . . . . . . . . . .
3.1.5.1 Requirements for flexible robot based automation .
3.1.5.2 Hybrid automation cost assessment . . . . . . . . .
3.1.5.3 Markets for hybrid assembly systems . . . . . . . .
3.1.5.4 Hybrid assembly in automotive industry . . . . . .
3.1.6 Conclusion and outlook . . . . . . . . . . . . . . . . . . . . .
A new cell production assembly system with human-robot cooperation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.2 System configuration of a new cell production assembly system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.2.1 Operation support technology . . . . . . . . . . . .
3.2.2.2 Information support technology . . . . . . . . . . .
3.2.2.3 Safety management technology . . . . . . . . . . .
3.2.3 System performance evaluation . . . . . . . . . . . . . . . . .
3.2.4 Conclusions and future work . . . . . . . . . . . . . . . . . .
63
65
65
65
71
72
77
77
78
78
80
81
81
82
84
84
88
89
90
91
92
92
93
98
99
100
102
102
103
104
106
107
108
109
4 Risultati, lavori futuri, conclusioni
111
Bibliografia
123
E LENCO DELLE TABELLE
1.1
Esempio check list attributi di progetto e di produzione. . . . . . . . . .
3.1
3.2
3.3
3.4
Tact time comparison between human and robot. . . . . . . . . . . . . . 94
Output calculation for human, robot and hybrid system assembly. . . . 96
Cost comparison between human, robot and hybrid system assembly. . 96
NPV calculation for hybrid system assembly. . . . . . . . . . . . . . . . 96
vii
7
E LENCO DELLE FIGURE
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
1.10
1.11
1.12
1.13
1.14
1.15
1.16
1.17
1.18
1.19
1.20
1.21
1.22
1.23
1.24
1.25
1.26
1.27
1.28
1.29
1.30
1.31
1.32
Schema della funzione logistica. . . . . . . . . . . . . . .
Flessibilità/Integrazione come risposta alla variabilità.
Automazione integrata e flessibile. . . . . . . . . . . . .
Diagramma P-Q. . . . . . . . . . . . . . . . . . . . . . .
Famiglie tecnologiche. . . . . . . . . . . . . . . . . . . .
Esempio di sistema FMS. . . . . . . . . . . . . . . . . . .
Macchina per lavorazione assial-simmetrica. . . . . . .
Centro di lavorazione rettilinea ad asse verticale. . . . .
Centro di lavorazione del legno. . . . . . . . . . . . . . .
Sistemi di movimentazione. . . . . . . . . . . . . . . . .
Tipologia di magazzino utensili. . . . . . . . . . . . . . .
Console di comando/controllo locale. . . . . . . . . . .
Layout. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Esempio di scomposizione in gruppi costruttivi. . . . .
Centraggio. . . . . . . . . . . . . . . . . . . . . . . . . . .
Riconfigurazione di 3 pezzi in 1 solo pezzo. . . . . . . .
Linea di assemblaggio. . . . . . . . . . . . . . . . . . . .
Attrezzo portante. . . . . . . . . . . . . . . . . . . . . . .
Esempio di sezionamento del sistema di trasporto. . . .
Esempio di sezionamento del sistema di trasporto. . . .
Sistemi di trasporto . . . . . . . . . . . . . . . . . . . . .
Tipi di caricatore . . . . . . . . . . . . . . . . . . . . . . .
Vibroalimentatore orbitale . . . . . . . . . . . . . . . . .
Esempio di ordinamento . . . . . . . . . . . . . . . . . .
Materiale continuo. . . . . . . . . . . . . . . . . . . . . .
Struttura di una stazione di alimentazione dei pezzi. . .
Struttura di un tratto di sicurezza. . . . . . . . . . . . .
Disposizioni consigliate. . . . . . . . . . . . . . . . . . .
Schema di una stazione robotizzata. . . . . . . . . . . .
Aree di operatività. . . . . . . . . . . . . . . . . . . . . .
Isola di lavoro manuale in esecuzione deviata . . . . . .
Stazioni di controllo . . . . . . . . . . . . . . . . . . . . .
3.1
3.2
Influence factors on robot assistance. . . . . . . . . . . . . . . . . . . . . 60
Combination of a manual and an automatic work station. . . . . . . . . 61
ix
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
2
3
4
6
8
10
10
11
11
13
13
16
18
19
20
21
23
23
24
24
25
26
27
27
28
29
29
30
30
31
32
x
3.3
3.4
3.5
3.6
3.7
3.8
3.9
3.10
3.11
3.12
3.13
3.14
3.15
3.16
3.17
3.18
3.19
3.20
3.21
3.22
3.23
3.24
3.25
3.26
ELENCO DELLE FIGURE
3.27
3.28
3.29
3.30
3.31
3.32
3.33
3.34
3.35
3.36
3.37
3.38
3.39
3.40
3.41
Cockpit install with IAD. . . . . . . . . . . . . . . . . . . . . . . . . . . .
IPAD - a reliable approach for advanced material handling. . . . . . . .
Workplace sharing systems. . . . . . . . . . . . . . . . . . . . . . . . . .
Workplace and time sharing hybrid system. . . . . . . . . . . . . . . . .
Assist robot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cobots. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Robot system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Humanoid/anthropomorphic robotic. . . . . . . . . . . . . . . . . . . .
Setup of the PORTHOS robot system. . . . . . . . . . . . . . . . . . . .
The mechatronic joint design of the LWR. . . . . . . . . . . . . . . . . .
Safety control - from the classical robot cell to human-robot interaction.
Safety controller for robot assisted welding. . . . . . . . . . . . . . . . .
Application of active opto-electronic protection devices in assembly. .
SafetyEYE camera for workplace surveillance. . . . . . . . . . . . . . .
Stereo camera based surveillance of human-robot-cooperation. . . . . .
Detection of a persons intruding arm. . . . . . . . . . . . . . . . . . . .
Impact tests performed by DLR and Fraunhofer IPA. . . . . . . . . . . .
Resulting HIC36 values at varying impact velocities for all robots. . . .
Spectrum of today’s gripping systems. . . . . . . . . . . . . . . . . . . .
Anthropomorph hand . . . . . . . . . . . . . . . . . . . . . . . . . . . .
IPA hands I and II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Vibratory brush feeder for gentle feeding of fragile parts . . . . . . . .
Vibratory bowl feeder with active sensing reorientation . . . . . . . . .
Camera equipped belt feeder for feeding and picking arbitrarily oriented components. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Adept Flexfeeders for multi component feeding. . . . . . . . . . . . . .
Camera and force controlled gripper fingers. . . . . . . . . . . . . . . .
Cobotic system for feeding and positioning of large assembly parts. . .
Essential results of the investigation. . . . . . . . . . . . . . . . . . . . .
Reduction of robot system costs. . . . . . . . . . . . . . . . . . . . . . .
Tact time diagram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cost reduction by use of intelligent assist devices (IAD). . . . . . . . . .
First admittance display prototype in industry. . . . . . . . . . . . . . .
Application of cobotic devices. . . . . . . . . . . . . . . . . . . . . . . .
New cell production. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mobile manipulators and parts kitting. . . . . . . . . . . . . . . . . . . .
Assembly operation with assistant of manipulator. . . . . . . . . . . . .
Software structure of skill transfer. . . . . . . . . . . . . . . . . . . . . .
Safety management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Difference in assembly time for experts and novices. . . . . . . . . . . .
61
62
63
64
66
67
68
69
70
71
72
73
73
74
75
76
76
77
78
79
79
83
83
4.1
4.2
Techniques for flexible feeding. . . . . . . . . . . . . . . . . . . . . . . . 112
Techniques for flexible feeding. . . . . . . . . . . . . . . . . . . . . . . . 117
87
88
89
92
93
94
95
98
98
101
103
104
106
106
108
109
Capitolo 1
L OGISTICA INTEGRATA E FLESSIBILE
’ OBIETTIVO di questo capitolo è quello di fornire una breve descrizione dei
sistemi logistici integrati e flessibili, che siano in grado di realizzare l’integrazione dei flussi fisici e dei flussi informativi per garantire: sia un elevato livello qualitativo dei prodotti e del servizio ai clienti, tramite la riduzione del tempo
di risposta all’ordine e il contenimento dei costi di produzione; sia di far fronte
con adeguata flessibilità al cambiamento continuo della gamma produttiva, conseguente alla variabilità e alla personalizzazione dei gusti del consumatore (per
una trattazione dettagliata dell’argomento Vedere [34]).
L
1.1
Definizione della funzione logistica
La logistica è la disciplina che tratta in maniera organica e sistematica la gestione
integrata (Figura 1.1) dell’intero ciclo operativo dell’azienda, industriale o del
terziario, attraverso le sue principali funzioni di:
- gestione dei materiali;
- gestione della produzione;
- gestione della distribuzione fisica dei prodotti finiti.
Figura 1.1: Schema della funzione logistica.
1
2
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
Si pone l’obiettivo fondamentale di garantire un elevato livello di servizio ai
clienti, fornendo prodotti di alta qualità, con rapidi tempi di risposta e a costi
contenuti. Gli strumenti per l’ottenimento di questi obiettivi sono:
À l’integrazione dei flussi fisici e informativi;
Á la flessibilità dei mezzi produttivi e logistici.
Al fine di far fronte con rapidità e senza oneri di adattamento al continuo
cambiamento della gamma produttiva conseguente alla variabilità del mercato
bisogna crescere nell’integrazione fra sistema informativo e sistema produttivo
attraverso:
L’integrazione : con una crescita modulare affinché i moduli siano interfacciabili
dal punto di vista software, hardware e meccanico;
La flessibilità : deve essere possibile riconfigurare i moduli mantenendo inalterata l’interfaccia con gli altri moduli.
1.2
Flessibilità come risposta alla variabilità
La variabilità del mercato odierno richiede una continua evoluzione ed innovazione dei tipi di prodotto offerti ed una loro sempre più spinta personalizzazione. Per questo bisogna realizzare una produzione flessibile ed elastica attraverso l’automazione flessibile dei sistemi e dei processi produttivi.
Figura 1.2: Flessibilità/Integrazione come risposta alla variabilità.
Il concetto di fabbrica automatica flessibile si è sviluppato come risposta alla
tendenza mondiale verso un tipo di produzione per lotti, tendenza indotta dalla
economia concorrenziale o di mercato. Questo ha portato:
- alla ricerca di nuove tecnologie di fabbricazione per la riduzione dei prezzi
dei prodotti;
1.2. FLESSIBILITÀ COME RISPOSTA ALLA VARIABILITÀ
3
- ad un incremento della varietà dei prodotti al fine di una maggior personalizzazione;
- all’introduzione frequente di nuovi prodotti al fine di ridurne il ciclo di vita;
- alla riduzione dei lotti di produzione.
Con l’inizio del nuovo secolo è iniziata l’era della produzione flessibile (fabbrica moderna), concomitante con il tramonto della produzione di massa (fabbrica tradizionale). La fabbrica moderna si caratterizza per:
À elasticità del sistema produttivo per rispondere a variazioni quantitative
del mercato;
Á flessibilità del processo produttivo per rispondere a variazioni qualitative
del mercato ovvero possibilità di variare il proprio mix produttivo. Questo
si traduce in:
- flessibilità nelle attrezzature di produzione;
- flessibilità nel sistema logistico;
- flessibilità nel sistema di gestione e controllo.
L’automazione integrata e flessibile è quindi lo strumento essenziale della
fabbrica moderna attraverso:
- l’automazione del flusso dei materiali e delle attività di produzione;
- l’automazione del controllo di processo.
Figura 1.3: Automazione integrata e flessibile.
Le attività di interazione e di elaborazione sono svolte da un programma che
utilizza come ingresso delle informazioni ottenute da un sistema di sensori e che
produce come uscita una modifica generata mediante un sistema di attuatori.
4
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
1.3
Analisi P-Q e tipologie di layout
La disposizione planimetrica o layout delle macchine e delle attrezzature costituenti l’impianto tecnologico principale dipende innanzitutto dal tipo di prodotto
(e quindi di processo da eseguire) e dal numero di prodotti da realizzare nell’unità di tempo (o potenzialità produttiva).
La scelta e la tipologia dei mezzi logistici per la movimentazione dei materiali
dall’inizio alla fine del processo produttivo dipendono a loro volta, oltre che
dalle caratteristiche fisiche e dimensionali delle materie prime, dai materiali in
lavorazione e dai prodotti finiti, dalle caratteristiche del layout.
Per procedere allo studio del layout generale di un sistema di produzione è necessario innanzitutto raccogliere informazioni sulla gamma dei prodotti Pk e sulle
rispettive potenzialità Qk (unità/anno) con k = 1, 2, . . . , n. Si realizza un diagramma P-Q (Figura 1.4) in cui i prodotti Pk sono rappresentati sull’asse delle
ascisse per potenzialità decrescente.
Figura 1.4: Diagramma P-Q.
I prodotti che ricadono nella zona I sono eseguiti in grande quantità, quindi favoriscono i metodi di produzione di massa attraverso linee di produzione dedicate. Si ottengono pertanto disposizioni di layout per prodotto nelle quali la collocazione delle macchine e dei mezzi produttivi rispecchia la successione delle
operazioni del ciclo tecnologico del prodotto.
I prodotti che ricadono nella zona II sono eseguiti in piccole quantità, quindi per
questi non è economicamente fattibile la costruzione di linee dedicate a ciascuno
di essi. Risultano quindi favoriti i metodi di produzione per reparti di lavorazioni
omogenee o a punto fisso, si ottengono in questo caso:
1. layout per processo: le lavorazioni vengono eseguite in reparti caratterizzati
da lavorazioni omogenee;
1.4. GROUP TECHNOLOGY E I SISTEMI FLESSIBILI DI FABBRICAZIONE
5
2. layout a posizione fissa: il prodotto rimane fermo al centro del layout e sono
gli operatori ed i mezzi operativi che si muovono intorno ad esso, facendo confluire i componenti e le parti necessarie (tipico per prodotti molto
voluminosi o pesanti).
Nelle zona centrale III si ha una numerosità intermedia della gamma produttiva
con corrispondenti valori intermedi di potenzialità. La risposta tradizionale è la
configurazione mista del layout ove vengono collocate linee produttive per i pochi
prodotti a potenzialità più elevata o addirittura piccole linee che realizzano una
parte del processo produttivo comune all’intera gamma, all’interno di un layout
per reparti.
La risposta innovativa è quella di ricorrere all’aggregazione di più prodotti simili
in famiglie. In questo modo la numerosità della famiglia diventa tale da giustificare una linea dedicata alla famiglia di prodotti con ottenimento del così detto
layout per famiglia di prodotti (Group Technology o tecnologia per famiglie). Necessariamente la linea deve essere flessibile in quanto lavora prodotti simili ma non
uguali.
Con riferimento alla Figura 1.4, l’analisi Prodotto - Quantità porta alla individuazione delle seguenti tipologie di layout:


 per
per

 per
1.4
0
Q>Q
layout per prodotto
00
Q<Q
layout per processo o a posizione fissa
00
0
Q < Q < Q layout misto o metodologia Group Technology
Group technology e i sistemi flessibili di
fabbricazione
Qualora la produzione sia caratterizzata da grande varietà di prodotto ma piccola
numerosità dei lotti, l’organizzazione per reparti è quella che garantisce la massima flessibilità. Per contro si hanno complicazioni nella gestione (soprattutto nel
flusso delle parti e delle informazioni) nonché flow time elevati.
1.4.1
La group technology
La Group Technology, avvalendosi del progresso tecnologico dei mezzi produttivi, consente di contemperare i vantaggi delle linee dedicate e delle organizzazioni flessibili. Ciò è reso possibile raggruppando prodotti “simili” dal punto
6
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
di vista produttivo e/o progettuale in famiglie. In questo modo si ottengono valori di numerosità dei pezzi della famiglia tali da rendere economicamente valida
l’introduzione di una linea (cella flessibile) dedicata alla famiglia stessa.
Figura 1.5: Famiglie tecnologiche.
Nel ricercare le similitudini si possono considerare principalmente::
• attributi di progetto (forma geometrica, dimensioni, etc.);
• attributi di fabbricazione (ciclo di produzione, attrezzature necessarie, etc.).
Nella pratica industriale per la generazione delle famiglie sono impiegate diverse metodologie.
Ispezione visuale
É questo il metodo più economico, semplice e rapido. La classificazione delle
parti in famiglie è fondata semplicemente sull’osservazione diretta dei prodotti,
eventualmente dei loro progetti (qualora non siano ancora in produzione) e/o
prototipi. É pertanto un metodo soggettivo e richiede una ottima conoscenza dei
processi di fabbricazione e dei prodotti.
Classificazione delle parti
É una metodologia maggiormente oggettiva rispetto alla precedente, consiste nella classificazione dei prodotti e delle loro parti secondo caratteristiche costruttive e/o funzionali. Nella sostanza si sviluppa l’analisi attraverso “check list”,
che permettono la formalizzazione delle principali caratteristiche di progetto e di
produzione, di cui la Tabella 1.1 ne riporta un esempio.
Analisi del flusso produttivo (productive flow analysis - PFA)
É il metodo più dispendioso e di conseguenza più preciso e oggettivo. Consiste nell’analizzare il ciclo di lavorazione piuttosto che il disegno delle parti
1.4. GROUP TECHNOLOGY E I SISTEMI FLESSIBILI DI FABBRICAZIONE
7
Tabella 1.1: Esempio check list attributi di progetto e di produzione.
Attributi di progettazione
Attributi di produzione
Forma esterna fondamentale
Forma interna fondamentale
Forma prismatica o rotazionale
Rapporto lunghezza/diametro (solidi di rivoluzione)
Rapporto d’aspetto (parti prismatiche)
Tipo di materiale
Funzione della parte
Dimensioni principali
Dimensioni secondarie
Tolleranze
Finitura superficiale
Processi principali
Operazioni secondarie
Sequenza delle operazioni
Macchina utensile
Utensili
Tempo di ciclo
Dimensione del lotto
Dimensioni principali
Attrezzature richieste
Tolleranze
Finitura superficiale
stesse al fine di individuare dei gruppi di prodotti realizzabili con gli stessi mezzi
produttivi. La procedura consiste in:
1. raccolta dei dati tramite “from-to-chart” e fogli operativi mono e multiprodotto;
2. ordinamento dei cicli di produzione;
3. costruzione delle carte PFA che rappresentano la descrizione dei cicli dei
vari prodotti;
4. raggruppamento delle lavorazioni (clustering).
Nella realtà industriale odierna le aziende sono portate dalla variabilità del
mercato alla produzione di ampi mix di prodotti. Questo fattore contrasta con
l’esigenza di raggruppare i particolari in famiglie (se si sceglie di impostare la
produzione seguendo i dettami della Group Technology). Spesso si è quindi chiamati a ricercare difficili compromessi tra queste due esigenze attraverso attività
di:
- razionalizzazione del progetto;
- razionalizzazione del processo ovvero del ciclo tecnologico;
- aggiornamento, semplificazione della gamma produttiva.
Attività di razionalizzazione del progetto
Essa consiste nella razionalizzazione, standardizzazione e modularizzazione dei
progetti e dei disegni costruttivi, tendente alla riduzione del numero dei particolari e alla unificazione nell’ambito dei rimanenti.
Razionalizzazione e standardizzazione dei cicli tecnologici
Tendente ad elaborare un ciclo standard per ogni famiglia di pezzi con riferimento ad un pezzo caratteristico (facente parte della famiglia o virtuale) il cui ciclo
8
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
tecnologico sia comprensivo di tutte le operazioni dei cicli di tutti gli altri pezzi
della famiglia a parità di sequenza operativa e di metodi di produzione. Per i singoli pezzi da realizzare variano solo i dati numerici del ciclo e possono mancare
al più alcune operazioni rispetto a quelle previste dal ciclo tecnologico del pezzo
caratteristico1 .
1.4.2
FMS - Flexible Manufacturing System
I sistemi FMS rappresentano tutt’oggi la risposta più soddisfacente per quanto
riguarda la fabbricazione di ampi mix di prodotti in piccoli lotti.
Sono generalmente costituiti da celle di macchine operatrici a controllo numerico
computerizzato servite da sistemi automatici di movimentazione dei pezzi e di
movimentazione degli utensili in grado di eseguire una gamma di differenti operazioni su materie prime o semilavorati.
Tali celle di fabbricazioni sono dislocate in linea e sono servite da un sistema
di movimentazione dei materiali, ad esempio un convogliatore o un sistema di
carrelli automatici, sul quale si muove il mix produttivo. Il trasferimento logistico dei vari pezzi da una cella all’altra ed il funzionamento delle celle medesime
viene gestito attraverso una opportuna rete gerarchica di calcolatori (Figura 1.6).
Figura 1.6: Esempio di sistema FMS.
1
Prodotto caratteristico: prodotto il cui ciclo produttivo comprende i cicli produttivi di altri
prodotti.
1.4. GROUP TECHNOLOGY E I SISTEMI FLESSIBILI DI FABBRICAZIONE
9
I principali vantaggi conseguiti all’introduzione di un FMS sono:
- massima flessibilità nel modificare l’output produttivo, per adeguare lo
stesso il più rapidamente possibile alle richieste del mercato;
- riduzione dei magazzini di reparto dei pezzi grezzi e semilavorati;
- riduzione dei costi di manodopera diretta ed indiretta;
- riduzione dei tempi di sosta del materiale in produzione;
- riduzione di attrezzature, maschere e utensili;
- ottimizzazione del carico macchina e del flusso produttivo;
- riduzione dei tempi di produzione;
- possibilità di lavorare su tre turni, anche in maniera non presidiata;
- riduzione dello spazio di layout occupato;
- miglioramento della qualità, particolarmente in termini di costanza.
É chiaro che la flessibilità si paga in termini di potenzialità produttiva e di
costo di investimento. Relativamente al diagramma P-Q illustrato in precedenza
(Figura 1.4), i sistemi FMS sono una risposta al problema della produzione relativamente alla zona III, mentre nella zona I (alte potenzialità e mix ridotto) il loro
uso non è appropriato.
1.4.3
Elementi costitutivi di un sistema FMS
In generale un sistema FMS è costituito da:
- macchine operatrici che realizzano le lavorazioni sui pezzi;
- sistemi di alimentazione/scarico dei pezzi alle macchine e di movimentazione
dei pezzi fra le macchine;
- sistemi per la gestione degli utensili;
- logiche di controllo per l’integrazione dei differenti dispositivi.
1.4.3.1
Macchine operatrici
Le macchine operatrici rappresentano il cuore del sistema, sono fondamentalmente riconducibili a 3 grandi famiglie di attrezzature:
1. macchine che eseguono lavorazioni assial-simmetriche: possono essere monomandrino o multimandrino. Possiedono un magazzino utensili con elevato
numero di vani (anche 200). Per applicazioni di piccole e medie dimensioni
sono preferibili quelle ad asse orizzontale, per le grandi dimensioni quelle
ad asse verticale (Figura 1.7).
2. macchine per lavorazioni lungo le coordinate, detti centri di lavorazione
rettilinea (Figura 1.8): sono deputate all’asportazione di materiale con moti
10
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
relativi rettilinei; dispongono di un magazzino utensili multiplo; le ultime
realizzazioni controllano fino a 12 assi in simultanea consentendo quindi
lavorazioni anche molto complesse.
3. macchine speciali, con questo termine si fa riferimento a tutte quelle macchine non inquadrabili nelle due precedenti categorie (Figura 1.9).
Figura 1.7: Macchina per lavorazione assial-simmetrica.
Figura 1.8: Centro di lavorazione rettilinea ad asse verticale.
1.4.3.2
Sistemi di movimentazione dei pezzi
All’interno di un sistema FMS i pezzi fluiscono in modo automatico. Per garantire questo flusso una rilevanza particolare è assunta dai sistemi di sostegno dei
pezzi ovvero i pallet portapezzo e le attrezzature. I pallet possono avere una duplice
funzione:
- sostenere ed ordinare i pezzi durante la fase di trasporto da una macchina
all’altra (tipico dei centri di lavorazione assial-simmetrici);
1.4. GROUP TECHNOLOGY E I SISTEMI FLESSIBILI DI FABBRICAZIONE 11
Figura 1.9: Centro di lavorazione del legno.
- sostenere ed ordinare i pezzi sia nella fase di trasporto ma anche in quella
di lavorazione in macchina (tipico dei centri di lavorazione lungo le coordinate).
Il pallet può essere montato in macchina secondo concezioni alternative (Figura
1.10):
Figura 1.10: Sistemi di movimentazione.
- alimentazione a tavola pendolare: la tavola portapallet della macchina esegue un moto pendolare per portare sulle slitte di lavoro un pallet con pezzo
12
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
-
-
grezzo. In tempo mascherato (si lavora sulla tavola pendolare mentre la
macchina esegue il ciclo) si provvede all’evacuazione del pezzo lavorato e
alla sua sostituzione con un nuovo grezzo;
alimentazione mediante shuttle: una navetta rotante a due posizioni scambia con la tavola della macchina il grezzo con il finito;
alimentazione mediante robot caricatore: un robot si occupa dello scambio
pallet (nei casi di centri di lavorazione assialsimmetrica);
alimentazione con magazzino locale a carosello: a bordo macchina è previsto un carosello che realizza uno stoccaggio di un certo numero di pallet:
questo può permettere il lavoro non presidiato per un certo intervallo di
tempo;
alimentazione di linea: variante rispetto allo schema a carosello, prevede
l’installazione dei centri di lavoro direttamente sulla linea di trasporto del
materiale.
Tipologie strutturali di pallet: pallet universali e pallet dedicati
Pallet universale: può essere adeguato allo specifico pezzo con staffagli e attrezzature mobili.
Pallet dedicato: ottimizzato nella geometria per il pezzo per renderne più veloce
l’impiego nella fase di bloccaggio/sbloccaggio.
La scelta dipende da una valutazione economica principalmente legata alla numerosità dei lotti e delle lavorazioni. In questi ultimi anni il mercato tende lievemente a preferire le attrezzature dedicate o personalizzate.
1.4.3.3
Sistemi per la gestione degli utensili
Le esigenze di automazione flessibile degli FMS impongono la sostituzione rapida e programmata degli utensili durante i cicli di lavoro. Gli utensili sono contenuti in magazzini disposti all’interno della macchina operatrice, questi magazzini locali hanno in genere configurazione a catena Figura 1.11a o a rack Figura
1.11b. Gli schemi operativi di sostituzione dell’utensile sono numerosi ma tutti
riconducibili all’uso di una navetta a doppia pinza che preleva dal magazzino
l’utensile per la nuova lavorazione, si avvicina al mandrino ne scarica l’utensile
in uso e monta quello per la lavorazione successiva. Fra i sistemi di controllo
di cui è equipaggiata una linea FMS rivestono grande importanza i sistemi di
rilevamento dell’usura e della rottura degli utensili.
1.4. GROUP TECHNOLOGY E I SISTEMI FLESSIBILI DI FABBRICAZIONE 13
(a) catena.
(b) rack.
Figura 1.11: Tipologia di magazzino utensili.
1.4.3.4
Logiche di controllo
Un sistema complesso come un FMS può funzionare correttamente solo se sostenuto da un adeguato sistema di gestione delle informazioni. La struttura modulare
del sistema ha favorito la sviluppo di una rete di controllo a schema gerarchico
tradizionalmente si individuano 4 livelli:
1. controllo locale del mezzo produttivo: costituito da computer industriale
o da PLC (Programmable Logic controller), è installato a bordo dei singoli
mezzi produttivi e ne gestisce la funzionalità (Figura 1.12);
Figura 1.12: Console di comando/controllo locale.
2. controllo di cella-sistema di alimentazione pezzi: costituito ancora da computer industriale o PLC, gestisce rispetto ai mezzi produttivi le risorse ausiliarie, quali:
- mezzi di movimentazione pezzi;
- stazioni di carico-scarico;
- eventuali stazioni di lavaggio.
14
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
3. DNC (Direct Numerical Control): il DNC rappresenta un sistema centralizzato per l’ottimizzazione della gestione in tempo reale del FMS. Il controllo DNC sovrintende tutto il funzionamento del sistema FMS, integrando il
controllo delle varie isole, dei sistemi di movimentazione intra-isola di tutte
le altre attività di linea;
4. CIM: il livello gerarchico più elevato è rappresentato dal CIM (Computer integrated Manufacturing): esso prevede la completa integrazione del
FMS e del relativo DNC con il sistema informativo aziendale. In particolare
il DNC può essere collegato con un host-computer di livello più elevato
che gestisce a livello centralizzato le varie attività aziendali coordinate intorno ad un unico Data Base. A questo livello di automazione comincia a
delinearsi nelle sue strutture fondamentali la fabbrica automatica.
1.4.4
Progettazione di un sistema FMS
La progettazione ed il corretto dimensionamento di un sistema FMS è fondamentale per la remunerazione dei notevoli investimenti necessari. Nel corso del
tempo si è andato consolidando un approccio basato su alcuni passi fondamentali:
• applicazione della Group Technology per la determinazione del mix di prodotti potenzialmente producibili mediante il sistema FMS;
• calcolo delle risorse necessarie basato su modelli a valori medi e relativa
determinazione della configurazione planimetrica;
• modellizzazione dinamica del sistema per la calibrazione dello stesso, in
modo da correggere eventuali problemi di congestioni dovuti al comportamento dinamico del sistema;
• valutazione economica della validità dell’iniziativa.
Applicazione della group technology
La flessibilità di un sistema FMS consente la realizzazione di mix molto ampi di
prodotti richiesti in piccole quantità. L’obiettivo di questa indagine è quello di individuare un certo numero di famiglie di prodotti simili, realizzabili con lo stesso
sistema flessibile di produzione, tali da rappresentare una numerosità di pezzi
così elevata da saturare la capacità produttiva del sistema e rendere vantaggioso
il costo al pezzo.
Calcolo delle risorse necessarie e disposizione planimetrica
L’obiettivo di questa fase è il calcolo del numero di risorse necessarie per la realizzazione delle famiglie di prodotti individuate. In secondo luogo risulta necessario studiare il problema della disposizione planimetrica delle risorse determinate.
1.4. GROUP TECHNOLOGY E I SISTEMI FLESSIBILI DI FABBRICAZIONE 15
Per quanto riguarda il calcolo del numero di macchine di lavoro, si tratta di determinare il parco macchine necessarie per produrre le quantità Qij (pezzi/anno)
di pezzi tipo j da realizzare sulla macchina tipo i. Dove con j = 1, 2, . . . , n si è indicato il mix di parti differenti da produrre e con i = 1, 2, . . . , m il set di macchine.
I pasi richiesti sono:
À studio dei cicli di lavorazione di ogni parte compresa nel mix;
Á calcolo del numero Z(i) =
dove:
Hd
H(i)
con:
ore effettivamente rese disponibili all’anno per la lavorazione di tutte le parti
j sulla macchina i
ore macchina all’anno relative ad ogni tipologia di macchina (ore/anno)
n X
t(i, j)
· Qij + T P M (i, j) · Nl (i, j)
CS (i, j)
H(i) =
in cui:
t(i, j)
Qij
CS (i, j)
T P M (i, j)
Nl (i, j)
A(i)
UT
DT
Cu
Co
H(i)
di macchine necessarie per il mix di parti.
Hd
j=1
A(i) · Cu · Co
tempo teorico di ciclo della macchina i per produrre la parte j (ore/pezzo);
produzione di pezzi/anno j richiesta sulla macchina i;
coefficiente di scarto della parte j lavorato sulla macchina i (pezzi
buoni/pezzi prodotti);
tempo teorico di preparazione (set-up) della macchina i per lavorare la parte
j;
numero previsto di lotti all’anno della parte j da lavorare sulla macchina i;
UT
coefficiente di disponibilità della macchina i
A(i) =
;
U T + DT
Up Time: tempo in cui la macchina è in grado di eseguire la propria funzione;
Down Time: tempo di fuori servizio dovuto a manutenzione, guasti, etc.;
coefficiente di conduzione: tiene conto del rendimento dell’operatore,
delle perdite di tempo per mancanza di motivazioni, condizioni di lavoro
sfavorevoli, condizioni ambientali non gradite, etc. (< 1);
coefficiente organizzativo, tiene conto della inefficienza organizzativa della
produzione (< 1);
Per quanto riguarda le altre risorse (magazzino, stazioni carico-scarico, etc.)
si tratta di scelte strettamente legate al caso reale, da valutare di volta in volta. Il
progetto del sistema di trasporto è strettamente legato alla disposizione planimetrica delle attrezzature di lavoro. Differenti sono le modalità con cui si possono
disporre le celle di lavoro:
16
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
- disposizione random: adatto per basse numerosità di macchine (3, 4), in genere
porta a linee di movimentazione dei pezzi lunghe e tortuose (Figura 1.13a);
- disposizione funzionale: le macchine sono raggruppate per funzione, in pratica è l’automatizzazione del concetto del job-shop (Figura 1.13b);
- disposizione modulare: i mezzi sono strutturali in celle di lavoro (di solito uguali) impiegate in parallelo (consente l’espansione modulare) (Figura
1.13c);
- disposizione cellulare: è la disposizione che meglio si adatta al concetto di
Group Technology, ciascuna cellula elementare contiene una o un gruppo
di macchine “particolarmente dedicate” alla realizzazione di una famiglia
(Figura 1.13d).
(a) Random.
(b) Funzionale.
(c) Modulare.
(d) Cellulare.
Figura 1.13: Layout.
Per ciascuna di queste disposizioni il flusso dei materiali è legato alla posizione reciproca delle macchine. In questo senso un’accurata analisi può portare
a dei notevoli risparmi economici e all’eliminazione di problemi di congestione
del traffico. Per questo studio si può adottare l’approccio Hollier (Vedere [10]).
Il metodo prende le mosse dalla conoscenza del flusso dei materiali fra le macchine della cella, quindi dalla conoscenza delle From-To chart (Foglio Da/A). Sommando le quantità per riga e per colonna si ottengono rispettivamente il numero
di pezzi uscenti dalla macchina in esame (Da) ed il numero di pezzi entranti nella
macchina in esame (A). Quindi per ciascuna macchina è possibile calcolare il rapporto Da/A. L’ordine delle macchine può essere stabilito per valori decrescenti
di tale rapporto (si mettono prima le macchine a flusso prevalente uscente). Una
volta determinata la disposizione delle attrezzature si può procedere alla scelta
del sistema di trasporto (non ci sono metodologie generali).
1.5. MEZZI LOGISTICI NELL’ASSEMBLAGGIO AUTOMATICO
17
Modellizzazione dinamica del sistema
Un FMS è un sistema molto complesso, governato da un numero molto elevato di
variabili che hanno caratteristiche di dinamicità molto forte. Dopo aver effettuato
il progetto di massima è assolutamente necessaria una valutazione del comportamento dinamico del sistema. Questo si può condurre attraverso la costruzione
di modelli simulativi che permettono la verifica dei principali dimensionamenti
effettuati alla luce dei transitori e delle congestioni del sistema.
Serve per:
- verificare se è raggiunto il livello produttivo richiesto sul mix;
- controllare la saturazione di tutte le risorse;
- correggere eventuali problemi introdotti dal fattore “tempo” sul sistema.
Valutazione economica dell’iniziativa
Alla fine di uno studio di carattere tecnico, vista l’entità delle risorse economiche
in gioco, è assolutamente necessaria la valutazione economica dell’iniziativa. Si
tratta di costruire il piano iniziale di investimento e provvedere alla stima dei
flussi di cassa annui alla luce dei dati ottenuti dal progetto (Vedere [47]).
1.5
Mezzi logistici nell’assemblaggio automatico
Molti prodotti vengono realizzati attraverso attività di fabbricazione di parti e
successivo assemblaggio (o montaggio) di queste. Anche i sistemi di montaggio
dovranno possedere un livello di automazione e flessibilità correlato alle quantità
e all’ampiezza del mix produttivo richiesto dal mercato.
1.5.1
Complessità del prodotto - struttura modulare
Un elemento rilevante nel processo di montaggio è la complessità del prodotto in
quanto il livello di difficoltà del montaggio è ad essa proporzionale. In base alla
complessità si distinguono diverse stazioni di lavoro:
- montaggio a posto singolo: grazie a un numero ridotto di elementi i prodotti
vengono montati completamente in posti di lavoro singoli;
- montaggio suddiviso: in presenza di prodotti complessi con un elevato numero di componenti, conviene frazionare i processi di montaggio.
Un presupposto dello smembramento dei prodotti complessi è costituito dal
fatto che il prodotto sia strutturato in modo tale da permettere il montaggio
preliminare di cosiddetti “gruppi costruttivi” (Figura 1.14).
18
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
Figura 1.14: Esempio di scomposizione in gruppi costruttivi.
Per una strutturazione conforme al montaggio dei gruppi costruttivi e dei
prodotti finiti è necessario osservare le seguenti regole fondamentali:
- l’intero processo di montaggio deve essere suddiviso in settori o “gruppi
costruttivi”;
- l’intero processo di montaggio deve essere chiuso e completo in se stesso, in
modo che la sua manipolazione corrisponda a quella di un singolo pezzo;
- un gruppo costruttivo deve poter essere collaudato separatamente;
- ciascun gruppo costruttivo deve presentare il minor numero possibile di
collegamenti con altri gruppi;
- i gruppi costruttivi che cambiano in funzione delle varianti del prodotto
non devono essere raggruppati insieme a quelli uguali per tutte le varianti;
- i gruppi costruttivi delle varianti devono presentare le stesse condizioni di
montaggio.
Per l’osservanza di queste regole è di importanza fondamentale la creazione
della cosiddetta unità di base per i singoli gruppi costruttivi e per il prodotto finito.
1.5. MEZZI LOGISTICI NELL’ASSEMBLAGGIO AUTOMATICO
1.5.2
19
Unità di base
Con unità di base si intende il singolo pezzo di base su cui nel corso del processo
di montaggio vengono poi montati gli altri elementi componenti. Essa deve essere strutturata in modo da rendere possibile il trasferimento da una stazione di
montaggio all’altra senza necessità di attrezzi particolari.
Questa esigenza non può essere soddisfatta in tutti i casi; tuttavia almeno per
quanto riguarda il montaggio manuale in cui si lavora senza attrezzi, tale condizione deve essere considerata come obiettivo. Nel caso delle catene di montaggio, con attrezzi portanti o nel montaggio automatizzato con supporti dei pezzi,
la funzione di unità di base può essere in parte svolta dagli attrezzi portanti o di
supporto. Un esempio classico di unità di base nell’elettronica è il circuito stampato su cui vengono poi montati tutti gli altri componenti.
Nel caso di montaggio con supporti dei pezzi, le unità di base devono essere progettate in modo da poter eseguire automaticamente il centraggio. La precisione
del centraggio dell’unità di base sul supporto è estremamente importante quando si esegue il montaggio suddiviso. Le piastre e gli alloggiamenti intesi come
unità di base devono essere configurati in modo che:
- centraggio esterno: le tolleranze esterne possono essere scelte consentendo
un posizionamento preciso nei punti di unione all’interno dell’unità di base
(Figura 1.15a);
- centraggio interno: nel caso di azione esterna possono essere precluse certe
attività lungo il perimetro e le superficie esterne del componente, l’unità
di base può essere bloccata internamente attraverso appositi fori (Figura
1.15b).
(a) esterno.
(b) interno.
Figura 1.15: Centraggio.
20
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
1.5.3
Numero di componenti
Direttamente collegato alla complessità del prodotto, il numero di componenti rappresenta un elemento critico per l’assemblabilità del prodotto finale. In
generale bisogna ottenere il miglior compromesso fra:
- la necessità di ridurre il numero di componenti per facilitare le operazioni
di montaggio;
- la necessità di limitare il ricorso a particolari unici ma costruttivamente
molto complessi e difficili da realizzare.
Con riferimento al montaggio l’uso dei materiali e dei metodi di produzione
moderni permette la configurazione strutturale in un solo pezzo di prodotti che
prima dovevano essere fabbricati assemblando più componenti (Figura 1.16).
Figura 1.16: Riconfigurazione di 3 pezzi in 1 solo pezzo.
1.5.4
Architetture di una linea di assemblaggio
Le linee di assemblaggio dedicate al montaggio dei componenti hanno 2 architetture principali:
1. Configurazione rettilinea: le stazioni di lavoro sono allineate unite tra loro
da un sistema di trasporto di tipo transfer. In relazione al numero di stazioni
e quindi agli ingombri si possono avere schemi a bastone, a C o ad anello
(Figura 1.17a);
2. Configurazione circolare: il sistema di trasferimento dei pezzi fra le stazioni
è rappresentato da una tavola rotante Figura 1.17b.
I particolari sono solidali a questa tavola che avanza di un passo ad intervalli di tempo, eseguendo una rotazione elementare e trasferendo i pezzi da
una stazione a quella seguente. Lo schema circolare a parità di stazioni di
lavoro ha:
- minore ingombro;
- schema più compatto.
1.5. MEZZI LOGISTICI NELL’ASSEMBLAGGIO AUTOMATICO
21
Peraltro non consente l’introduzione di un numero molto elevato di stazioni.
Per attenuare questo limite si sono introdotte le tavole multigiro Figura 1.17c.
In questo caso il pezzo viene completato con un numero di giri della tavola
superiore ad 1. Quindi la flessibilità dei mezzi operativi in una determinata
posizione angolare è un prerequisito imprescindibile, in quanto in relazione
al numero del giro in questione devono eseguire un differente lavoro sul
particolare.
(a) rettilinea.
(b) circolare.
(c) circolare multigiro.
Figura 1.17: Linea di assemblaggio.
Si definisce cadenza di una stazione di lavoro il tempo fra l’uscita di due prodotti consecutivi dalla stazione stessa. Le linee di assemblaggio possono essere
progettate:
22
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
1. a cadenza imposta: tutte le stazioni di lavoro hanno la medesima cadenza e
quindi lo stesso tempo di ciclo per eseguire un’operazione:
- utilizzate quando vi è bassa incidenza di operazioni manuali;
- utilizzate con schema circolare;
2. a cadenza non imposta: le stazioni lavorano con tempi di ciclo differenti e
quindi sorge il problema della loro armonizzazione o bilanciamento attraverso l’inserimento di polmoni:
- utilizzate quando vi è alta incidenza di operazioni manuali;
- utilizzate con schema rettilineo.
1.5.5
Elementi costitutivi di una linea flessibile di assemblaggio
I principali elementi presenti in una linea flessibile di assemblaggio FAS (Flexible
Assembly System) sono:
-
attrezzi portanti del pezzo;
sistema di trasporto dei pezzi;
sistema di alimentazione dei componenti da montare;
stazioni di lavoro;
stazioni di controllo e collaudo dei montaggi.
1.5.5.1
Attrezzi portanti del pezzo
Il pezzo base deve essere trasferito fra le stazioni di lavoro in modo da “crescere”
tramite i successivi montaggi di parti componenti fino a diventare il prodotto
finito. Il pezzo base per poter essere trasferito e centrato viene riferito a un attrezzo portante (definito anche pallet, vassoio, supporto) che verrà movimentato dal
sistema di trasporto.
In alcuni casi il vassoio oltre al pezzo portante può anche contenere altre parti
del prodotto finito che poi nelle stazioni di lavoro verranno via via assemblate.
Si parla in questo caso di “kit” di componenti disposti sul pallet.
I vassoi possono contenere un pezzo base ovvero essere multipli cioè contenere
più pezzi base. Nel caso di vassoi multi-pezzo sono impiegabili 2 modalità di
lavoro:
- attraverso attrezzi portanti semplici con pezzi allineati in differenti posizioni
(Figura 1.18a);
- attraverso attrezzi portanti circolari con possibilità di movimento di rotazione in modo da presentare il pezzo da lavorare sempre nella stessa
posizione (Figura 1.18b).
1.5. MEZZI LOGISTICI NELL’ASSEMBLAGGIO AUTOMATICO
(a) con pezzi allinati.
23
(b) circolare.
Figura 1.18: Attrezzo portante.
1.5.5.2
Sistemi di trasporto degli attrezzi portanti
Trattando il tema dei sistemi di trasferimento degli attrezzi portanti implicitamente si fa riferimento a linee di assemblaggio rettilinee. Nel caso di inserimento
di postazioni manuali di lavoro, al sistema di trasporto è richiesta anche la capacità di realizzare il magazzino interoperazionale (necessario per svincolare i
diversi tempi di lavorazione).
Questo requisito può essere ottenuto mediante:
1. sezionamento del sistema di trasporto in tante unità elementari controllabili
separatamente da opportuni attuatori e sensori in modo da avere moto solo
in caso di possibilità di avanzamento del pezzo (Figura 1.19);
Figura 1.19: Esempio di sezionamento del sistema di trasporto.
2. impiego di attuatori con trasmissioni a frizione che al raggiungimento di un
certo sforzo disaccoppiano il sistema di trasporto dall’attuatore. In questi
sistemi il polmone si ottiene automaticamente: in quanto i pezzi si adagiano uno contro l’altro, essendo fermi fanno crescere lo sforzo delle motorizzazioni fino al livello di soglia, “sganciando” così le trasmissioni di moto
(Figura 1.20);
Uno sviluppo ulteriore di questa soluzione, prevede l’eliminazione della
trasmissione a frizione a favore di sistemi dissipativi: un’opportuna scelta
24
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
Figura 1.20: Esempio di sezionamento del sistema di trasporto.
dei materiali dei sistemi di trasporto permette di ottenere per attrito il moto
in condizioni di via libera e lo stazionamento in situazioni di immagazzinamento a causa della via occupata. Nella pratica industriale i sistemi più
impiegati sono:
-
a catene (Figura 1.21a);
a rulli (Figura 1.21b);
a nastro (Figura 1.21c);
a tapparelle (Figura 1.21d).
(a) a catena.
(b) a rulli.
(c) a nastro.
(d) a tapparella.
Figura 1.21: Sistemi di trasporto
1.5. MEZZI LOGISTICI NELL’ASSEMBLAGGIO AUTOMATICO
1.5.5.3
25
Sistemi di alimentazione dei componenti
Nelle stazioni di lavoro delle linee di assemblaggio vengono eseguiti dei montaggi fra componenti e gruppi costruttivi che possono essere direttamente posizionati sul pallet oppure possono provenire dall’esterno rispetto al vassoio stesso. Tendenzialmente i componenti di dimensioni ridotte oppure quelli variabili
da prodotto a prodotto vengono alimentati dall’esterno.
Nel caso di impiego di dispositivi automatici di montaggio è auspicabile per semplificare e velocizzare le operazioni che il materiale sia alimentato in modo ordinato al dispositivo di montaggio. I sistemi di alimentazione, detti anche caricatori,
si suddividono in:
- caricatori piani (Figura 1.22a);
- caricatori a contenitore (Figura 1.22b);
- vibroalimentatori orbitali.
(a) piano.
(b) contenitore.
Figura 1.22: Tipi di caricatore
Il dispositivo di alimentazione più diffuso nella pratica è il vibroalimentatore orbitale (Figura 1.23). Questo dispositivo è formato da un basamento pieno,
appoggiato su tre piedini di gomma che assorbono le vibrazioni. Sulla circonferenza di questo basamento sono disposti in modo obliquo tre pacchi di molle a
lamina che lo collegano con la lastra di supporto del contenitore dei pezzi. Nel
contenitore si trovano i canali elicoidali che conducono all’uscita del vibroalimentatore. Le vibrazioni sono create da un magnete alimentato a corrente alternata. Nei canali elicoidali sono sistemati gli elementi che servono per ordinare i
pezzi, chiamati anche deflettori o selezionatori di giaciture. Con il vibroalimentatore orbitale i pezzi possono essere depositati, ordinati e poi, in collegamento
con le guide di uscita, inseriti in caricatori e trasportati. Grazie alla forma convessa del fondo e alle oscillazioni meccaniche provocate dall’elettromagnete, i
pezzi disposti casualmente si spostano lungo il bordo del deposito. Il magnete
26
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
azionato da corrente alternata genera forze di attrazione periodiche, cosicché il
contenitore oscilla con una vibrazione prossima alla risonanza e in virtù della posizione inclinata delle molle durante la vibrazione si verifica un avanzamento dei
componenti.
(a) Schema.
(b) Esempio.
Figura 1.23: Vibroalimentatore orbitale
I pezzi vengono ordinati attraverso 2 sistemi:
1. ordinamento passivo: i pezzi in posizione scorretta vengono respinti e tornano a cadere nel contenitore di raccolta (Figura 1.24a);
2. ordinamento attivo: i pezzi vengono portati nella giusta posizione attraverso elementi di ordinamento disposti in serie e quindi non ricadono mai nel
contenitore (Figura 1.24b).
Un presupposto fondamentale per poter ordinare i pezzi è che questi possano
muoversi liberamente nell’area d’azione degli elementi di ordinamento, in modo
da non ostacolarsi reciprocamente. Inoltre inserendo degli elementi di separazione è possibile evitare un afflusso troppo elevato di pezzi da ordinare.
Materiale a nastro continuo unito a un supporto
Nella pratica industriale talvolta sono da alimentare componenti che presentano
delle forme “intricate” difficili da alimentare con i sistemi visti prima. In questi
casi il metodo più sicuro di alimentazione è l’alimentazione di materiale in forma
di striscia (Figura 1.25). Il componente viene alimentato come materiale continuo
e separato dalla striscia solo al momento del montaggio. Il vantaggio di questo
tipo di fornitura è il mantenimento dell’ordine dei pezzi e quindi l’ottenimento
di una disponibilità molto elevata.
1.5. MEZZI LOGISTICI NELL’ASSEMBLAGGIO AUTOMATICO
(a) passivo.
(b) attivo.
Figura 1.24: Esempio di ordinamento
Figura 1.25: Materiale continuo.
27
28
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
Schema di una stazione di alimentazione dei componenti
La Figura 1.26 mostra una tipica stazione di alimentazione dei componenti alle
stazioni di lavoro localizzate lungo una tavola di lavoro. Possiamo riconoscere:
-
un vibroalimentatore orbitale;
una guida elettromagnetica di uscita;
una stazione di singolarizzazione;
un manipolatore “pick and place”, che trasferisce il componente proveniente dal singolarizzatore e lo colloca sul supporto del transfer della macchina di montaggio.
Figura 1.26: Struttura di una stazione di alimentazione dei pezzi.
Il grado di affidabilità del sistema è determinato dalla:
- qualità dei componenti;
- pulizia dei pezzi;
- disposizione e la precisione delle stazioni di alimentazione.
L’esperienza pratica ha evidenziato che i guasti più frequenti si verificano all’interno del vibroalimentatore orbitale e nel passaggio da quest’ultimo alla guida di uscita. Per garantire che i guasti non compromettano la disponibilità della
macchina di montaggio, la lunghezza della guida di uscita deve essere dimensionata in modo da creare un effetto polmone tra i componenti ordinati, che si
trovano nella guida di uscita. La capacità di questo polmone deve essere determinata in modo da permettere che i guasti siano individuati ed eliminati tempestivamente senza che la guida giri a vuoto. A tale scopo conviene prevedere un
tratto di sicurezza nella guida di uscita, inteso come polmone, dotato di un opportuno sistema di rilevamento di segnali e di un corrispondente comando. La
Figura 1.27 mostra la struttura schematica di un tratto di sicurezza.
1.5. MEZZI LOGISTICI NELL’ASSEMBLAGGIO AUTOMATICO
29
Figura 1.27: Struttura di un tratto di sicurezza.
Sulla guida di uscita devono essere installati due sensori S1 e S2 :
• con il sensore S1 il vibroalimentatore orbitale viene adattato alle prestazioni
richieste. Se la guida di uscita scorre a pieno carico questa condizione viene
rilevata dal sensore S1 che disattiva il vibroalimentatore. Quando i pezzi
vengono scaricati nel singolarizzatore attraverso il manipolatore e l’ultimo
pezzo supera la stazione S1 lo stesso sensore riattiva il vibroalimentatore;
• un secondo sensore S2 viene disposto in modo tale che tra il sensore e la
stazione di singolarizzazione ci sia un tratto di sicurezza con una certa capacità di polmone. Se il sensore S2 registra che in quel punto non arrivano
più pezzi traduce questa condizione in un segnale di allarme ottico o acustico. In questo modo l’addetto alla macchina viene avvertito che nell’area
del vibroalimentatore si è verificato un problema.
Collocazione dei sistemi di alimentazione
In genere essi vengono collocati in modo ortogonale ai transfer rettilinei e in senso radiale sulle tavole circolari per limitare gli ingombri e soprattutto non precludere spazi per altri attrezzi (Figura 1.28). In alcuni casi però si è costretti ad
impiegare soluzioni ibride e non ottimali.
Figura 1.28: Disposizioni consigliate.
30
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
1.5.5.4
Stazioni di lavoro
Per quanto riguarda le stazioni ove effettivamente vengono realizzate le operazioni di montaggio si evidenzia una grande variabilità e differenziazione di
tipologie. Sul piano della funzionalità operativa una prima osservazione riguarda la configurazione della stazione, in cui vi possono essere:
- dispositivi completamente automatici;
- operatori umani.
In genere gli alimentatori vengono posti sia per gli operatori umani che per i
dispositivi automatici molto vicini alle zone di montaggio in modo da minimizzare i percorsi di presa e manipolazione delle parti, anche se con configurazioni
diverse (Figura 1.29).
Figura 1.29: Schema di una stazione robotizzata.
Per l’uomo vengono definite 2 aree di operatività (Figura 1.30):
- l’area dei movimenti primari (AP)
- l’area dei movimenti secondari (AS)
Figura 1.30: Aree di operatività.
Per eliminare il ricorso ad operazioni nell’area secondaria, vengono disposti
i sistemi di alimentazione frontalmente e entro i 35 cm di distanza. Stazioni di
1.5. MEZZI LOGISTICI NELL’ASSEMBLAGGIO AUTOMATICO
31
lavoro manuali sono tipiche in linee non cadenzate, quindi risulta necessario l’inserimento di magazzini interoperazionali che permettano il disaccoppiamento
delle cadenze di lavoro. La modalità più impiegata è la costruzione di un’area
di lavoro ricavata in esecuzione deviata rispetto al transfer con l’intatta possibilità della via di corsa rettilinea. Il tratto deviato di transfer offre la possibilità di
immagazzinare prodotti (Figura 1.31).
Figura 1.31: Isola di lavoro manuale in esecuzione deviata.
Altro aspetto rilevante per le stazioni di lavoro riguarda l’attrezzistica, che
dipende dall’operazione da svolgere di volta in volta. Per quanto riguarda invece
le postazioni di lavoro bisogna tenere in considerazione il tema della sicurezza.
1.5.5.5
Stazioni di controllo
Durante le operazioni di montaggio l’unità in via di costruzione viene sottoposta
a dei controlli per verificare la correttezza dello stato del Work In Progress (WIP).
I risultati positivi provocano il “via libera” mentre i risultati negativi possono
essere interpretati nei seguenti modi:
- disattivazione immediata: in questo caso qualora la macchina rilevi un errore essa procede all’immediato arresto della linea o comunque dei mezzi
responsabili di quel montaggio;
- disattivazione della macchina dopo un certo numero di errori in sequenza: in questo caso il mezzo operativo viene disattivato solo dopo un certo
numero di errori consecutivi;
- memorizzazione degli errori: questa modalità non prevede l’arresto della
stazione di montaggio. In caso di errore il pezzo errato viene semplicemente reso riconoscibile per i dispositivi a valle che non lo utilizzeranno
nelle successive fasi di montaggio.
Per quanto riguarda il tipo di struttura, le stazioni di controllo possono essere
suddivise tra:
- stazioni che a causa della posizione dell’oggetto da esaminare, devono essere disposte in posizione fissa (Figura 1.32a);
- stazioni che, per svolgere i propri compiti, devono eseguire un movimento
(Figura 1.32b).
32
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
(a) fissa.
(b) mobile.
Figura 1.32: Stazioni di controllo
1.5.6
Progettazione di una linea flessibile di assemblaggio
automatico
La progettazione di una linea di montaggio automatico (FAS) richiede una attività piuttosto complessa, che, pur differenziandosi da prodotto a prodotto si può
ricondurre allo sviluppo di alcune fasi tipiche.
1.5.6.1
Razionalizzazione del progetto dei prodotti secondo il Design For
Assembling - DFA
Per ottenere un agevole montaggio automatico di un prodotto occorre che il progetto del prodotto stesso abbia delle caratteristiche intrinseche di semplicità che
favoriscano le attività di montaggio. Quindi in fase di progetto bisogna:
1. effettuare delle scelte che semplifichino il ciclo di montaggio riducendo
il numero delle parti nonché il numero e la difficoltà delle operazioni di
montaggio da eseguire (si parla infatti di Design For Assembling - DFA);
2. applicare criteri di modularità e di semplificazione dei prodotti;
3. che le linee vengano raggruppate per famiglie di prodotti (Group Technology)
In letteratura si trovano tabelle che formalizzano alcuni principi progettuali
per l’ottenimento di prodotti facili da montare (Vedere [2]). Per una analisi sistematica si può far ricorso alla cosiddetta “DFA house” introdotta da Rampersad
(Vedere [48]).
1.5.6.2
Razionalizzazione del processo produttivo
Consiste nel progettare ed approntare processi di assemblaggio realizzabili e tecnologicamente simili per prodotti della stessa famiglia. In questo caso si pos-
1.5. MEZZI LOGISTICI NELL’ASSEMBLAGGIO AUTOMATICO
33
sono cercare delle razionalizzazioni attraverso la modularità e la semplificazione
dei cicli di assemblaggio.
1.5.6.3
Definizione del prodotto caratteristico
Consiste nell’individuazione del prodotto caratteristico, ossia un prodotto reale
o fittizio, che presenta tutte le lavorazioni di tutti i prodotti della famiglia. In
altre parole tutti i prodotti della famiglia si devono poter ottenere da quello
caratteristico per eliminazione di fasi di lavoro (montaggio).
1.5.6.4
Determinazione dei tempi e dei metodi di montaggio
Per poter dimensionare le stazioni di lavoro bisogna definire le metodologie di
montaggio e i tempi corrispondenti. La valutazione dei tempi può essere condotta sperimentalmente attraverso campagne di rilevazione sul campo di processi
analoghi e successive elaborazioni statistiche dei risultati. La stima dei tempi di
montaggio può essere svolta per via teorica attraverso la procedura Method Time
Measurement - MTM. In base a questa procedura ciascuna azione di montaggio
può essere scomposta in operazioni elementari a cui corrisponde un tempo standard. Sommando tutti i tempi elementari si ottiene una stima della durata della
particolare attività di assemblaggio (Vedere [40]).
1.5.6.5
Scelta della cadenza della linea e dimensionamento delle stazioni
Le linee di assemblaggio possono lavorare con una cadenza imposta oppure libera. La scelta del tipo di cadenza viene attuata in relazione alla produttività
richiesta ed alla componente di lavoro umano necessario per le particolari attività di montaggio previste.
In generale si ha:
- cadenza imposta in presenza di linee con elevata produttività e ridotta incidenza del lavoro umano;
- cadenza libera in presenza di numerose attività richieste all’operatore.
Scelta la cadenza bisogna distribuire il carico di lavoro in un certo numero di
stazioni. L’obiettivo di questo bilanciamento è quello di ottenere il miglior compromesso economico fra costi diretti di montaggio (manodopera ed attrezzature)
e i costi indiretti di mancato completamento di alcune operazioni, nel rispetto
dei vincoli tecnologici e produttivi presenti. Nel caso di linee a cadenza imposta
viene impiegato il metodo euristico di Kottas Lau. Nel caso di linee a cadenza
libera si esegue una ricerca del bilanciamento che riduce al minimo l’ammontare
degli ozi per i mezzi produttivi.
34
CAPITOLO 1. LOGISTICA INTEGRATA E FLESSIBILE
Nella pratica industriale il parametro di valutazione del bilanciamento ottimale
è dato dal minimo costo unitario di montaggio.
1.5.6.6
Scelta del sistema di trasporto e di alimentazione dei componenti
Si passa ora alla scelta del sistema di trasporto per la movimentazione dei pezzi
e/o degli attrezzi portanti lungo le varie posizioni di montaggio. La presenza di
stazioni manuali implica il dimensionamento dei polmoni fra le diverse stazioni
(in pratica realizzati sul sistema di trasporto). Quindi a questo stadio della progettazione si effettua la scelta del tipo e della collocazione fisica dell’insieme dei
caricatori e alimentatori. In pratica il risultato di questa fase è il progetto di
dettaglio delle strutture fisiche della linea attraverso disegni, tabelle, schemi, etc.
1.5.6.7
Verifica della flessibilità e simulazione
Prima di procedere alla valutazione economica dell’investimento è bene quindi
verificare se la linea ideata rispetta le richieste in termini di flessibilità. In pratica
si tratta di controllare la “montabilità” di tutti gli elementi della famiglia. Questo
controllo può sanare eventuali lievi imprecisioni introdotte dalle possibili differenze tra il progetto concettuale e le attrezzature fisiche realmente scelte. Inoltre
è consigliabile un approccio simulativo per la verifica del comportamento dinamico in condizioni di transitorio, per la determinazione della capacità richiesta
dai polmoni per poter cogliere gli obiettivi produttivi prefissati.
1.5.6.8
Valutazione economica dell’investimento
La linea di assemblaggio è fattibile se presenta una adeguata convenienza economica a questo scopo alla fine del progetto bisogna determinare, attraverso la
valutazione dei costi di installazione e di esercizio delle attrezzature scelte, il ritorno economico dell’investimento (attraverso il calcolo del VAN, TIR, tempo di
recupero) e il costo unitario stimato di montaggio.
Capitolo 2
F LEXIBLE AUTOMATION STATE OF
THE ART
T
HE DEMAND for customized products has been increasing continuously in the
recent years and a great deal of attention has been given to the automation
of manufacturing and assembly systems. In order to meet increased demand for
customized products and to reduce production lot sizes, the industry has adapted
new techniques and production concepts by introducing flexibility into production machines (2.1) so that variety of products can be manufactured on the same
equipment.
A key obstacle to the economic application of flexible automation is parts feeding (2.2). Traditionally, dedicated devices such as vibratory bowl feeders have
performed parts feeding. Such devices have a high cost, and are dedicated to
a single component geometry and, consequently, the number of dedicated feeders required for a planned variety of products is equal to the number of distinct
part types. This results in a high capital cost that can only be justified in a dedicated mass production environment, where there are long production runs and
few product changes. Dedicated feeding methods are also usually too inflexible
for robotic assembly systems where there is a wide range of component types in
small batches being produced at lower rates.
2.1
2.1.1
Flexible systems
A new cell production assembly system with
human-robot cooperation - 2010
In manufacturing, where the wages of operators are high and the availability of
expert operators remains limited, it is proving impossible to improve the efficiency of cell production assembly systems. Even novice operators are asked to
35
36
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
achieve high levels of productivity and reliability with a diverse range of products. To satisfy this demand, the authors ([8]) have developed a new cell production assembly system with human-robot cooperation. This system consists of
three key technologies; parts feeding by double manipulators on a mobile base,
production process information support for the operator, and safety management
for cooperation between the operator and the robot. The safety management
for the cooperation is new from the viewpoint that industrial robots can cooperate with human operators in automatic operation without stopping, which can
achieve higher productivity. The authors demonstrated that, for the assembly
of a cable harness, productivity with the new cell can be doubled relative to the
conventional manual cell production. Furthermore, the authors proved that the
assembly quality with the new cell production method can also be greatly improved because assembly errors can be almost entirely eliminated. In addition,
the new method of automatic parts kitting and feeding by the mobile manipulators allows the operators to concentrate on the assembly operations and improve productivity in comparison with conventional manual parts kitting and
feeding. This new cell production assembly system is expected to be installed
in manufacturing factories in the near future. To promote the installation, the
authors will make greater efforts to reduce the cost of the setup including the
mobile manipulators and the safety management, and strive to improve safety
even further.
2.1.2
Bicriteria robotic operation allocation in a flexible
manufacturing cell - 2010
Consider a manufacturing cell of two identical CNC machines and a material
handling robot. Identical parts requesting the completion of a number of operations are to be produced in a cyclic scheduling environment through a flow shop
type setting. The existing studies in the literature overlook the flexibility of the
CNC machines by assuming that both the allocation of the operations to the machines as well as their respective processing times are fixed. Consequently, the
provided results may be either suboptimal or valid under unnecessarily limiting
assumptions for a flexible manufacturing cell. The allocations of the operations
to the two machines and the processing time of an operation on a machine can be
changed by altering the machining conditions of that machine such as the speed
and the feed rate in a CNC turning machine. Such flexibilities constitute the point
of origin of the current study. The allocation of the operations to the machines
and the machining conditions of the machines affect the processing times which,
in turn, affect the cycle time. On the other hand, the machining conditions also
affect the manufacturing cost. This study ([13]) is the first to consider a bicriteria model which determines the allocation of the operations to the machines, the
processing times of the operations on the machines, and the robot move sequence
2.1. FLEXIBLE SYSTEMS
37
that jointly minimize the cycle time and the total manufacturing cost. The authors
provide algorithms for the two 1-unit cycles and test their efficiency in terms of
the solution quality and the computation time by a wide range of experiments
on varying design parameters. The authors have formulated the problem as a
Mixed Integer Nonlinear Programming (MINLP) model. the authors developed
an exact solution procedure for the S1 cycle. Since the allocation problem for the
S2 cycle is NP-Complete, they presented a heuristic algorithm that generates a
set of approximate efficient solutions. The authors compared the results of the
algorithm with commercial MINLP solvers DICOPT and BARON. The proposed
algorithm proved to be very efficient in terms of the number and the quality of
the generated solutions and the computational requirements.
2.1.3
An analysis of cyclic scheduling problems in robot
centered cells - 2010
The focus of this study ([3]) is a robot centered cell consisting of m computer
numerical control (CNC) machines producing identical parts. The existing robotic cell scheduling literature mainly focuses on in-line or mobile robotic cells. In
many practical applications, robot centered cells are used simply because they
require less space than in-line robotic cell layouts. Furthermore, stationary base
robots (as in robot-centered cells) are cheaper to install and easier to program
and, consequently, more robust than mobile robots. Two pure cycles are singled
out and further investigated as prominent cycles in minimizing the cycle time.
It has been shown that these two cycles jointly dominate the rest of the pure cycles for a wide range of processing time values. For the remaining region, the
worst case performances of these pure cycles are established. The special case
of 3-machines is studied extensively in order to provide further insight for the
more general case. The situation where the processing times are controllable is
analyzed. The proposed pure cycles also dominate the rest when the cycle time
and total manufacturing cost objectives are considered simultaneously from a bicriteria optimization point of view. Moreover, they also dominate all of the pure
cycles in in-line robotic cells. Finally, the efficient frontier of the 3-machine case
with controllable processing times is depicted as an example. Interestingly, pure
cycles are used extensively in metal cutting industry, not because they are provably optimal, but because they are very practical and easy to understand and
implement. More specifically, in a pure cycle, each part is loaded and unloaded only once, which means less gaging, one probable reason why this cycle is
preferred in practice. Future lines of research directions might be to extend the
current study to include multiple part types or dual gripper robots.
38
2.1.4
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
Pure cycles in flexible robotic cells - 2009
In this study ([41]), an m-machine robotic cell used for metal cutting operations is
considered. The machines used in such manufacturing cells are CNC machines
which are highly flexible. As a consequence, each part is assumed to be composed of a number of operations and each machine is assumed to be capable of
performing all of the required operations of each part. The authors investigated the productivity gain attained by the additional flexibility introduced by the
CNC machines. A new class of robot move cycles, namely the pure cycles, which
resulted from the flexibility of the machines are defined. The problem is formulated as a special travelling salesman problem (TSP), where the distance matrix
consists of decision variables as well as parameters. Due to the extensive computational effort required to solve this formulation, they determined two specific
pure cycles which perform effectively. They determined the regions of optimality
for both of these cycles. For the remaining small region, they determined worst
case bounds for both of these cycles. They also proved that the set of pure cycles dominates all flowshop-type robot move cycles. The results show that these
proposed cycles are not only simple and practical but perform very efficiently
as well. As a design problem, they considered the number of machines in a cell
as a decision variable and determined the optimal number of machines for the
two specific pure cycles. Extending the analysis to the multiple parts case can
be considered as a future research direction. In such a case, the determination
of the part input sequence must also be tackled which will certainly increase the
complexity of the problem even further.
2.1.5
Cooperation of human and machines in assembly lines 2009
The paper ([15]) gives a survey about forms of human-machine cooperation in assembly and available technologies that support the cooperation. The close cooperation of human and machine in hybrid assembly is motivated by the increased
need for flexibility, adaptability and reusability of assembly systems. Hybrid assembly systems can essentially reduce the amount of fixed production costs in relation to variable costs. The effectiveness of these systems depends on the lot size,
but also on the design of the cooperative workplace and its automated systems.
The overall effectiveness of hybrid assembly also depends on the intelligent feeding of workpieces to the cooperative workplace. An essential precondition for the
time efficient close cooperation between human and machine is the safety of the
worker. A lot of research has been carried out in this field in recent years and
first sensor systems for the surveillance of the interaction between human and
robots are available on the market. The research in the field of intelligent assist
devices (IAD) that efficiently support the worker is the basis to keep the work-
2.1. FLEXIBLE SYSTEMS
39
er in the loop in order to utilize his cognitive and senso-motoric advantages for
highly flexible assembly. The support functions of IADs comprise safety as well
as reduction of physical strain and efficient synchronisation between worker and
automated systems. Future research will have to focus on different aspects for
highly cooperative hybrid assembly. The robustness of the system parts for safety, load reduction and others will have to be increased in order to extend the field
of applications. A new challenge will be to improve the cooperation of human
and machine not only for a single human but also for a group of workers that
have to fulfil a common task supported by one or more machines at the same
time. The close cooperation between human and machine may in the future also
change the way that automated systems will be programmed. It can be foreseen, that a wide range of automated cinematic systems such as robots will in
the future include interfaces for efficient teach modes based on the direct force
coupled interaction between human and robot. The human oriented automation
approach may generally lead to a change in the design principles for robots. Today design of industrial robots is focused for accuracy. Control with application
of external sensors is needed for stable and safe interaction. In long term solutions the design of intrinsically safe robots will focus on safety and sophisticated
control will provide accuracy. By this, the next generation of robots will interact
with human directly for cooperative manipulation, where the robot is responsible
for load-bearing and precision whereas the human can contribute with sensing,
intelligence and skills. Generally the advantages of a close cooperation between
human and intelligent assist systems in assembly lines build a basis for a more
flexible and also more sustainable way of assembly and disassembly in the future.
2.1.6
A market approach to decentralized control of a
manufacturing cell - 2009
With manufacturing system evolves from traditional centralized organization to
more and more decentralized federation, effective coordination of the resource
allocation becomes more and more important. For balancing the influences of various conflicting managerial objectives with an unified instrument in scheduling
practice, market approach seems to be an appealing choice. This paper ([6]) introduces a multi-agent framework with bidding mechanism for dynamic scheduling
of a CNC manufacturing cell consisting of multiple non-identical workstations.
The market model developed is based on a heuristic periodic scheduling procedure the authors proposed. A pricing policy is designed for the effective mutual
selection between jobs and workstations. Simulation results show the advantage
of theirs proposal over some other approaches. Further observations reveal that
as workload arises and job due date tightness increases, the necessity also increases for gathering more jobs arrived over a longer time horizon to get more
reasonable resource allocation. This observation gives us some idea on how to
40
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
select between the policy of periodic bidding and the policy of continuous bidding. Future lines of research directions might be to extend the current study to
finding an effective way for the precise determination of the appropriate length
of the scheduling horizon T under a certain given workshop configuration.
2.1.7
Modeling of machine failures in a flexible manufacturing
cell with two machines served by a robot - 2008
In this study ([38]), a stochastic model is developed to analyze performance measures of a flexible manufacturing cell (FMC) under different operational conditions, including machine failures and repairs. The FMC consists of two machines
served by a robot for loading and loading purposes, and a pallet handling system.
Stochastic models (Markov processes) and the closed-form solution formulas obtained in this paper could be used to analyze and optimize the productivity and
other performance measures of an FMC under different machine, robot and pallet operational characteristics. Using the models presented in this paper, best
parameter combinations can be determined for a given FMC system. In particular, best machining rates, robot loading and unloading rates, pallet capacity and
pallet transfer rates can be determined for a given set of FMC machine characteristics. Furthermore, reliability and availability analysis of the FMC system can
be determined based on different failure/repair characteristics of the machines in
the system. It is possible to optimize machine repair rates, based on other system
parameters, to achieve maximum production output rates and other performance
measures.
2.1.8
Effects of maintenance policies on the productivity of
flexible manufacturing cells - 2006
This research ([26]) was undertaken to determine the effects of various maintenance policies on the operational capability (production output rate and availability) of flexible manufacturing cells (FMC). Flexible manufacturing cells are operated at higher usage rates than the traditional equipment since they are flexible
and can allow manufacturing of a wide variety of parts. Therefore, they are vulnerable to more wear and tear during their useful life. Maintenance is considered
extremely important under such conditions. However, no detailed study can be
found in the literature on the effects of maintenance policies on the operational
condition of FMC. Five distinct maintenance policies were identified and their
effects on production rate, which is a direct outcome of availability, are analyzed
by using mathematical formulation of failure rates and simulation modeling. Using SIMAN simulation package, six simulation programs were developed( one
for the fully reliable FMC and one for each of the five maintenance policies implemented on the FMC). The results of the analysis of several experiments show
2.1. FLEXIBLE SYSTEMS
41
that maintenance of any form has significant effect on production output rate or
the availability of the FMC. However, the type of maintenance applied is important and should be carefully studied before implementation. The implication of
this research is that any FMC system under consideration must be analyzed with
respect to several maintenance policies and the best policy should be selected before blindly implementing a policy. As it is seen from the analysis above, the best
policy in all cases appears to be opportunity-triggered maintenance policy (OTP)
and the worst policy is the corrective maintenance policy (CMP) case, which is
operate-to-failure and then repairs case. One important consideration that is not
incorporated into the present study is related to the cost aspects. Cost data were
not available during the course of this research. Future studies can be carried out
on the cost aspects of various policies if such data are available. The best cost saving policy can be determined depending on the specified parameters related to
the repair costs and the preventive maintenance costs. In order to perform such
a study, one has to collect all maintenance related costs for the system under consideration. Other possible maintenance policies must be studied and compared
to those presented in this study. Combinations of several policies are also possible in the same FMC system. For example, while a machine is maintained by one
policy, another machine could be maintained by a different policy. These aspects
of the problem need further investigation.
2.1.9
Design configuration for a mixed-model assembly system
in case of low product demand - 2006
This paper ([17]) addresses the application of a mixed-model assembly balancing problem to an assembly-to-order environment in the case of low production
rates and large number of tasks. The aim of this work is to propose an alternative
design procedure for the balancing of semi-automated and mixedmodel assembly systems under low product demand effects by the application of multi-turn
circular transfers, such as a multi-stations rotating table. This layout configuration permits a job enlargement for human operators and, at the same time,
provides an increment in task repeatability through the work-pieces assembling
by increasing the number of the turns of the transfer. Finally, the developed
heuristic procedure is tested on a simple rotating table assembly cell, a partial
representation of a complete assembly system of domestic air compressors.
2.1.10
Dynamic scheduling in flexible assembly system based
on timed Petri nets model - 2005
This paper ([24]) investigates a scheduling model for optimal production sequencing in a flexible assembly system. The system features a set of machines
working together in the same workspace, with each machine performing a sub-
42
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
set of operations. Three constraints are considered: (1) the precedence relation
among the operations specified by the assembly tree; (2) working space that limits concurrent operations; and (3) the variation of process time. The objective is
to find both a feasible assignment of operations to machines and schedule tasks in
order to minimize the completion time for a single product or a batch of products.
The assembly process is modeled using timed Petri nets and task scheduling is
solved with a dynamic programming algorithm. The method calculates the time
required precisely. A detailed case study is discussed to show the effectiveness
of the model and algorithm. Future work is required to implement the optimization method with multiple variants to enhance the reliability and robustness of
the algorithm. This approach of modeling, planning and control for an FMS will
also be further developed.
2.1.11
A stochastic model for the analysis of a two-machine
flexible manufacturing cell - 2005
This paper ([11]) presents a stochastic model to determine the performance of a
flexible manufacturing cell (FMC) under variable operational conditions, including random machining times, random loading and unloading times, and random
pallet transfer times. The FMC under study consists of two machines, pallet handling system, and a loading/unloading robot. After delivering the blanks by the
pallet to the cell, the robot loads the first machine followed by the second. Unloading of a part starts with the machine that finishes its part first, followed by
the next machine. When the machining of all parts on the pallet is completed,
the handling system moves the pallet with finished parts out and brings in a new
pallet with blanks. A model with these characteristics turns out to be a Markov
chain with a transition matrix of size 5n + 3, where n is the number of parts on
the pallet. In this paper, the authors present exact numerical solutions and economic analysis to evaluate FMC systems, to determine optimal pallet capacity
and robot speed that minimize total FMC cost per unit of production. Design
engineers and operation managers can benefit from these analysis either during
the FMC design phase for the appropriate selection of its components or during
its operation phase. The stochastic model of the described FMC can be extended
in many directions. A new problem emerges if the number of machines in the
FMC is increased to three or more machines instead of two. It will be attractive if
the model of this problem turns out to be a Markov chain with an exact solution.
Another related problem arises when we assume that each machine is served by
a dedicated robot. Although the robot utilization would to be less in this case,
it may be interesting to witness the effect on the production rate and machines’
utilization and compare them to our model. A further extension is to study and
analyze an FMC with unreliable machines and/or robot.
2.1. FLEXIBLE SYSTEMS
2.1.12
43
Image based 3D Surveillance for flexible
Man-Robot-Cooperation - 2005
In today’s industrial production men and robots usually work separately from
each other in order to avoid accidents. By this, the advantages of an interoperable automated and manual assembly can not be used for a flexible and efficient
production. If it would be possible to overcome the separation of man and robot,
the accuracy and speed of robots could be combined with the flexibility and reliability of human workers. Especially for complex assembly and handling tasks,
this combination is useful. A system based on digital 3D image analysis has been
developed, which supervises the common working area of robot and man. The
digital image processing succeeds more and more in replacing conventional safety devices in an automated production. The main advantages are the economical
hardware platform and the possibility of a flexible safety area definition. The
introduced surveillance technology ([32]) is based on image processing methods
for motion analysis, skin colour detection, three-dimensional positioning with a
trinocular sensor arrangement and utilisation of human body geometry for defining critical areas. By this it is possible to determine 3D coordinates of humans and
machines respectively. The motion areas of men and robot must no longer be partitioned from each other and a safe overlapping operation can take place. Thus
the special capability characteristics of humans and machines can be combined
in a co-operative work place which complies with the increasing demands for
flexible assembly resulting from a more and more individualised production.
2.1.13
A heuristic approach to batching and scheduling a single
machine to minimize setup costs - 2004
In this paper ([5]) the authors analyzed the problem of creating batches of jobs
and scheduling tools necessary for their processing. This problem arises typically
in the management of flexible manufacturing systems. Here, they considered the
special case of a single processing facility (e.g. robot) with degree of parallelism
k = 3, in which no more than k parts may be accommodated in the cell at the
same time. In particular, we restricted our research to the case in which exactly
k = 3 parts are always present in the system.
The objective is the maximization of the productivity of the cell, which corresponds to considering the minimization of the time by which all the jobs are
completed (make span). Since the number of batches b is fixed (due to the above
assumption), the operations processing times are given and setup times (tools
and batches setup times) are independent from the sequence and identical for all
the jobs, the objective easily translates into the minimization of the total number
of tool setups. Since even for k = 3 the problem is shown to be hard (Agnetis et
al. 2003), they developed a heuristic solution algorithm, based on LCS concept,
44
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
that has proved to be effective (in terms of solution values) and efficient (in term
of computation times) in practice.
A possible direction for further research, strictly linked to the solution scheme
here proposed, is the development of more sophisticated heuristic algorithms
(e.g. based on local search meta-heuristic concept such as tabu search or simulated annealing) able to further reduce the optimality gap. A different possibility
is the development of approximation algorithms which, differently from heuristic ones, guarantee the solution quality a priori, even in cases when no optimal
solution is available.
2.1.14
Framework for designing a flexible cellular assembly
system - 2004
The aim of the present paper ([29]) is to develop a conceptual framework for the
simultaneous design and control of a flexible, agile reconfigurable and robust assembly system in conjunction with the analysis and optimization of product, process, system structure, material-handling devices and plant layout. An effective
solution for reconfigurability and agility in assembly to order and batch-type systems is a modular semi-automatic approach based on cellular flexible facilities:
modularity, automation and human skills are combined to gain the advantages
of mass production. Families of parts are produced in flexible cells, i.e. groups
of various machines and resources that are physically close together and process
one or more families of similar parts. The proposed approach is based on a multistage iterative procedure that integrates different supporting decision techniques
and tools such as design for assembly, group technologies and cellular manufacturing. An application to the optimization of a semi-automatic flexible assembly
system is illustrated: the system performances of example alternative solutions
are presented and compared.
2.1.15
Mini assembly cell for the assembly of mini-sized
planetary gearheads - 2004
This paper ([36]) discusses a novel assembly method and system for a commercially available 8mm diameter miniature planetary gearhead. The system comprises a commercially available four-degree of freedom industrial robot, two vision systems, a force feedback system in the robot wrist, and specially designed
flexible part feeders. The system has proved successful in assembling planetary gear units independently. Depending on the task, one can select whether
to use the high accuracy and repeatability of the robot or alternatively use programmable frequency vibration in the gripper to stochastically align the parts
that the robot handles.
2.1. FLEXIBLE SYSTEMS
2.1.16
45
An approach to the design of a manufacturing cell under
economic considerations - 2002
In this case study ([4]), a new method to finalize the design of a manufacturing
cell aiming for its profit maximization was proposed. The proposed method integrates simulation techniques, design of experiments, regression analysis, Taguchi
methods, and a new profit model to facilitate the generation and evaluation of
different designs. The application of the method was demonstrated with a case
study. The case study included optimal setting of number of associates; raw material inventory policy (frequency, order size, and safety stock); number of machines for a specific operation; and number of setups per planning period. The
method was used to generate several designs with high estimated average profit values and low sensitivity to changes in selected non-controllable economical
factors. The generation of these designs allowed a DM to choose the best design
according to the profit value, robustness, and other practical considerations. The
final evaluation of the resulting designs suggested that TM is a viable and simple way to generate a cell design with high profit values and low variability to
noise factors. It also became evident that regression analysis can be used to generate a wide span of designs with tradeoffs between them that can be assessed
easily with the proposed profit-based method. One advantage of the regression
approach is that it produces a number of high-profit designs for the DM to select
from while TM produces just one design. In conclusion, the results of this study
suggest that the design of a manufacturing cell can be carried out by using potential profitability of the system as the central objective.
Future research
It is recommended that the following ideas be examined in the future:
1. The problem described is not exclusive to manufacturing cells. The proposed method could be used to approach the design of different manufacturing systems;
2. After finishing the full factorial enumeration, the number of alternatives to
be chosen for further study is an arbitrary number. It could be researched
as to how to determine the number of alternatives that must be selected to
raise the probability that the optimal solution is among the selected set;
3. The setting of factors at levels such that the average profit is maximized and
the variability around this average is minimized can be found by solving the
optimization problem with more efficient means than the total enumeration
used here.
46
2.1.17
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
Knowledge intensive Petri net framework for concurrent
intelligent design of automatic assembly systems - 2001
To effectively carry out the design of assembly systems, an effective approach
needs to be developed. However, due to the complexity of design process, various assumptions or simplifications have been introduced in the existing methods
to obtain solutions for the problem. The problem using the existing methods to
design assembly system is that the unmanageable complexity could occur both
in computation cost and storage. The complexity and limitations using the existing methods can be greatly alleviated with the observation as well as idea behind
the work reported in this research. This paper ([31]) presented a systematic and
concurrent intelligent approach to the modeling, design, and planning of flexible
robotic assembly systems. There are two major ideas employed in this approach.
One is the use of knowledge Petri net to establish a number of place and transition template. Specific FASs can thus be derived by instantiation from this set
of templates. The other is the use of function-behavior-structure model for design and evaluation of FASs. The knowledge Petri net is an enhancement of the
conventional Petri nets by employing AI knowledge representation techniques
including rule-based technique and frame. It allows the established behavior
analysis techniques for Petri nets to be carried out. At the same time, the system complexity can be reduced since any FAS model which built upon a given
set of place and transition template can be made more compact. Therefore, the
knowledge Petri nets can build a unified, complete framework to represent multiview knowledge and to perform reasoning and learning. Compared to the existing methods, the work essentially has an integrative nature, and the proposed
knowledge-based function-behavior-structure design model and framework are
more generic. They can be used as an alternative to reduce the computing cost
and storage in the design of assembly system. The framework was implemented
based on the knowledge Petri net models combined with POSES++ and the conventional procedural language. It has potentials that support all development
stages including general design, detailed design, programming, simulation, validation and verification in a consistent and integrated manner. The developed integrated design environment is a 3D modeling, animation, and simulation package, allowing the user to intuitively and interactively build and animate 3D models. The examples show that the approach is feasible, and is an effective means
to achieve concurrent design of FASs. The results obtained from this work are
generic enough to be applicable for other types of FMS such as flexible machining
centers, flexible assembly systems, semiconductor manufacturing systems, etc.
However, the work presented in this paper is just a preliminary effort in the automatic assembly system design. A large amount of work to make the proposed
approach and system practical use is desired to be done. One of the future efforts
will be required to further the research on the mapping from function model to
2.2. TECHNIQUES FOR FLEXIBLE FEEDING
47
structure for the conceptual design of assembly system. In the meanwhile, the
further development of the system is also required.
2.2
2.2.1
Techniques for flexible feeding
Development and testing of a brush feeder - 2010
In this paper ([20]) a brush feeding system based on tilted brushes is presented. It allows the controlled motion and the correct alignment of products different in shape, weight and surface roughness. The brush feeder is analyzed both
from a theoretical and experimental point of view. The brush feeder demonstrated good positioning and aligning capabilities both in passive configuration
(the chute) and in the active one (vibrating feeder). Several configurations of
the feeder have been successfully tested with different components. The paper
presents some experimental evidences of the achieved results. The analysis presented here focused on the most influencing parameters, but other parameters
and feeder characteristics will be object of future investigations both theoretical
and empirical.
2.2.2
Design of a modular feeder for optimal operating
performance - 2010
This paper ([18]) presents a design method for a new modular feeder, which ensures an optimal operational behaviour at single or double line frequency through
geometric adaptation. The method is based on a structural mechanical model of
the system, which was derived from the Lagrangian Equations. Additionally, the
best ways for receiving the required model parameters were determined. The
model and the adaptation method were verified with a product neutral basic
set-up. Occurring deviations were relatively small and therefore can be equalled
with an amplitude controller. Finally, the method was included in the design process of the modular feeder, which now has to be tested on industrial applications.
Further research should focus on the theoretic determination of the micro-slip
effect to avoid the measurement of the leaf spring stiffness.
2.2.3
Approximation to the dynamics of transported parts in a
vibratory bowl feeder - 2009
In this paper ([12]) vibration in an electromagnetic drive unit, of the kind typically used for vibratory part feeders, is modelled. The results obtained were used
to analyse the dynamic behaviour of a part moving along the spiral track of a
vibratory bowl feeder. This analysis led to a series of analytical results covering
dynamic, geometric and electromagnetic parameters. What marks this research
48
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
as different from previous works is the analytical evaluation of the effects of the
electromagnetic parameters and associated geometric parameters. The main interest in the results obtained here is the possibility of using analytical results in
order to predict part behaviour and select a feeder adapted to particular needs,
as well as to optimise the design of a vibratory bowl feeder and its drive unit. As
a demonstration of this potential, the authors numerically simulated the effects
of certain of the parameters featuring in the expressions that resulted from theirs
analysis. Likewise, in order to reinforce the consistency of the numerical results,
the behaviour of a part was simulated, using the visual Nastran program, for
two different bowl friction coefficients, obtaining results that were entirely congruent. Future lines of research will include a study of the analytical resolution
of the complete equation and the verification of these results by real tests.
2.2.4
On-line dimensional measurement of small components
on the eyeglasses assembly line - 2008
In this paper ([44]), a novel on-line measuring system for the inspection of small
metallic subassemblies for the eye-glasses industry is presented. The automated system proposed is based on artificial vision, and exploits two CCD cameras
and an anthropomorphic robot to inspect and manipulate the subcomponents
of the eyeglass. Each component is recognized by the first camera in a quite
large workspace, picked up by the robot and placed in the small vision field of
the second camera which performs the measurement process. Finally, the part
is palletized by the robot. The system can be easily taught by the operator by
simply placing the template object in the vision field of the measurement camera (for dimensional data acquisition) and hence by instructing the robot via the
Teaching Control Pendant within the vision field of the first camera (for pick-up
transformation acquisition).The major problem dealt with is that the shape and
dimensions of the subassemblies can vary in a quite wide range, but different
positioning of the same component can look very similar one to another. For this
reason, a specific shape recognition procedure was developed. In the paper, the
whole system is presented together with first experimental lab results.
Further testing is currently being performed with the aim of optimizing and
expanding the system. Future developments will include the integration of a
flexible bulk part feeder in our device, to obtain a completely autonomous inspection system. Another benefit of adding this option will be the possibility
of teaching the recognition, measure and manipulation of a reduced number of
poses for each subassembly (at least one single pose for each part, possibly the
more probable one). In this way, the time required to add a new component to
the database will be drastically reduced. During the workcycle, after picking the
recognized objects (i.e. the ones which dropped in the taught poses), the objects
2.2. TECHNIQUES FOR FLEXIBLE FEEDING
49
holding the unknown poses will be flipped by the feeder; finally, a new image of
the recognition subsystem will be acquired.
2.2.5
Dynamic modeling and experimental verification of a
piezoelectric part feeder in a structure with parallel
bimorph beams - 2007
The study ([23]) is aimed to perform dynamic modeling of a part feeder powered
by piezoelectric actuation. This part feeder consists mainly of a horizontal platform vibrated by a pair of parallel piezoelectric bimorph beams. It starts with
an establishment of the dynamic equations of motion of the piezo-beams in the
piezo-feeder via the Rayleigh-Ritz method and is then followed by the modeling
on the impact dynamics between the platform and transported part, which is accomplished via basic collision theory. The validity of the established models is
further ensured by experiments. The closeness shown between the theoreticallypredicted dynamics of the feeder/part and those experimental counterparts finally confirm the success of the modelings. Based on the analytical and experimental
results obtained, the following conclusions can be drawn.
1. The first three assumed-modes proposed are capable of predicting the realistic dynamics of a flexurally-vibrated piezoelectric beam, as evident from
the experimental results.
2. The basic collision theory is pertinent to be employed to predict the impact dynamics between the platform and transported part, leading to wellpredicted part transporting speeds.
3. It is theoretically predicted and experimentally confirmed that the transported part experiences three stages of different types of motions: (1) part
arising with the platform; (2) part free falling; and (3) part impacting the
platform. This series of interactive three-stage dynamics is repeated between platform oscillation cycles, moving the part from one end to another
of the platform. The transported part moves horizontally with the platform
in the period of arising with the platform, while it undergoes horizontal
motion in a contact velocity in periods of free-falling and collisions.
4. A few general rules for control strategy have been distilled:
• Based on experimental operation, the feeder should be operated under
the resonance frequency of 330 Hz to avoid structural break-down.
• A basic rule of maximizing the part traveling speed is to increase the
input voltage for a shorter period of part stiction to the platform.
50
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
Although the part-traveling speeds theoretically predicted in this study anticipate well about the realistics, a broader distribution range of experimental
part-traveling speeds as compared to the theoreticals is present. This might be
due to the facts that (1) the transported part is not a point mass and (2) the ignorance on the impact friction between the part and platform. To further improve
modeling, the part needs to be assumed as a rigid body and one needs to investigate the dynamic effects of the impact friction on the part-transporting speed in
the future.
2.2.6
Advanced vision guided robotics provide “future-proof”
flexible automation - 2006
Vision guided robotics (VGR) is a fast growing technology and a way to reduce
manpower and retain production, especially in countries with high manufacturing overheads and labour costs. The paper ([1]) describes the new automation
system of the Swedish company SVIA. Shows that the need to position components to a set pick-up position is eliminated - the vision system determining the
position of randomly fed products by a recycling conveyor system. The vision
system and control software gives the robot exact coordinates of the components,
which are spread out randomly beneath the camera field of vision, enabling the
robot arm to move to a selected component and pick from the conveyor belt.
2.2.7
Economic application of piezoelectric actuator in linear
vibratory feeding - 2006
The research ([25]) investigated a decoupled vibratory feeding scheme driven by
a piezoelectric actuator. In the decoupled design, the action of the horizontal
and vertical vibration would be separately controlled to achieve a desired track
trajectory and consequently the feed of the part. A computer control method
was used to develop voltage input waveforms to control these actuators to produce decoupled movements that cause parts to move forward relative to the track
and overcome backsliding and other inefficient feeding motions of conventional
operation. To provide a commercial result from research, the authors proposed
an economic application of decoupled feeding in an industrial environment, in
which a regular PZT bimorph bender is used to drive a linear vibratory feeder in
micro-industry. The bimorph actuator is economic and easy to install; however,
it needs to be properly controlled to obtain the desired actuation. In the present
work, a simple but effective control algorithm was designed to suppress undesired responses of the bimorph actuator, i.e. the disturbance from the harmonic
vibration and hysteresis effect. It was noticed that there are many uncertainties
inside the acceleration profile during practical feeding; therefore feeding speed
cannot be simply calculated, and many factors, such static and motion friction,
2.2. TECHNIQUES FOR FLEXIBLE FEEDING
51
track flatness, part add in and fall-out, should all be considered further to precisely predict feeding speed. However, it is shown that the control of a decoupled
vibratory feeder is feasible and the trend of speed varying based on controlled
parameters is clear. Therefore, it is practical and economical to use a commercial
PZT actuator in a high-precision-required condition with appropriate control.
2.2.8
A new type of parts feeder driven by bimorph piezo
actuator - 2005
A novel meander-line structure implemented with bimorph piezoelectric actuators driven by two sets of alternating current power with phase difference is
developed in this article ([9]). Via the generated traveling wave, this mechanism
is able to transport parts. The dynamic modeling of the structure, the driving control circuitry design, the motion trajectory analysis and the optimal transport feed
rate is studied, and also verified by the practical experiment. The meander-line
mechanism is expected to have the advantage of bearing heavier load as compared with the usually seen in-line type of parts feeder with the same size and
property of the bimorph actuator. To increase the displacement for better transport efficiency is feasible by changing the size and the property of the bimorph
actuator and the connector.
2.2.9
A modular contactless feeder for microparts - 2005
This paper ([7]) describes the development of a feeder for mini and micro parts.
A complete study of a feeding method based on electrostatic fields is presented.
Through a theoretical analysis and experimental results the optimized configuration of the feeder is proposed, shown and discussed. The performances of
possible alternative configurations are quantitatively compared according to the
variation of the process parameters. In addition the paper presents a flexible and
reconfigurable system based on standard modules easy to be linked by rapid set
up operations. A shorter switching period T can increase the feeding rate up to
a maximum which is different from part to part: beyond that value parts cannot
follow the travelling field. This actual limit of the feeder will be faced in the future development according the principle of parallel feeding through new shapes
of the electrodes.
2.2.10
Piezo actuated vibratory feeding with vibration control 2005
This research ([45]) investigated the effect of decoupling the motion of a conventional vibratory feeder in order to develop a one-axis vibratory sliding mode for
feeding tiny delicate parts that could ultimately be adapted to two axes. This
52
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
research investigated a new type of vibratory drive by employing a piezoelectric actuator. In a decoupled design, the action of the horizontal and vertical
vibration would be separately controlled to achieve a desired part trajectory and
consequently the feed rate of the part. A computer control scheme was used to
develop voltage input waveforms to control these actuators to produce movements that cause the parts to move forward relative to the track and overcome
backsliding and other inefficient feeding motions. There are two advantages to
this type of design. First, it can handle the parts in a more controlled manner by
keeping them in constant contact with the track while maintaining a high conveying speed. This results in good protection for fragile parts. Secondly, this
concept results in feedback control schemes that could allow the computer to
adjust the drive waveforms to optimize feeder performance. Due to the nonsinusoidal nature of driving waveforms, the higher harmonic components can
excite undesired deformation of the responding trajectory. A self-tuning control
based on dynamic system identification was employed to counteract the higher
harmonic effect. A one-dimensional hardware model is built and experiments
were conducted to verify the precision and feasibility of the controlled actuation.
The results are promising for industrial applications of decoupled vibratory parts
feeding using piezo-electric actuators.
2.2.11
Development of a flexible and programmable parts
feeding system - 2005
This paper ([21]) describes the development of a flexible and programmable vibratory bowl feeding system which is suitable for use in a flexible manufacturing
system. Controlled by computer and driven by electro-pneumatic cylinders and
stepper motors, this feeding system is capable of identifying the orientations of
on-rotational parts and actively re-orientating them into the desired orientation.
A neural network is incorporated into the system to identify the orientation of
parts moving on the track of the vibratory bowl. The neural network obtains
scanned signature of the surface of the parts and compares them with learned
patterns. With the use of optical sensors, internal features such as holes and pockets can be detected. Three types of neural network architectures (ARTMAP, ART2
and Back-propagation) were evaluated for their capability in pattern recognition
and classification of the feed orientation of parts in the vibratory bowl. ARTMAP
was found to yield the best results. The system developed extends the capability
of conventional bowl feeders to include feeding parts with only internal features,
and feeding a family of similar parts without costly re-tooling. These features are
well suited to applications in flexible manufacturing systems.
2.2. TECHNIQUES FOR FLEXIBLE FEEDING
2.2.12
53
Modal analysis and control of a bowl parts feeder
activated by piezoceramic actuators - 2004
Vibratory bowl feeders are the most versatile and common feeding and orientating devices for automatic assembly. Thus, they are being widely used in many
industry fields. The electromagnet has been and being commonly used as an exciting actuator in these vibratory bowl feeders. However, because of complexity
of its mechanical structure and limited capability, there exist various impending
problems such as severe noise, non-linear motion of parts, passive characteristics
and so forth. Despite of many research works to improve dynamic performance
of the conventional electromagnet actuator-based bowl parts feeder, there still exists some significant problems; noise and passive motion. As one of solutions to
resolve these problems is to use piezoactuators as a new exciting technology. The
piezoelectric material is one of smart materials that can develop mechanical strain
when subjected to an electric field, or, alternatively can develop an electric field
subjected to mechanical deformation. Advantages of the piezoelectric material
include fast response time, wide frequency bandwidth and accurate control capability. By adopting the piezoelectric material as an exciting actuator, we can devise more effective bowl parts feeder providing accurate and adjustable feeding
speed of parts than the conventional one activated by the electromagnet actuator.
In this work ([37]), as the first phase to develop highly effective piezoactuatordriven bowl parts feeder, a system model is established using a finite element
method (FEM). By adopting commercial software package, modal characteristics
of the proposed bowl parts feeder are analyzed and compared with experimentally measured ones. In addition, control aspects of the parts feeder are experimentally investigated by emphasizing the feeding speed of parts with respect to
the intensity of the input voltage. The results presented in this research work
are quite self explanatory justifying that the bowl parts feeder activated by the
piezoceramic actuators can be effectively used in the automatic assembly line.
It is finally remarked that optimal adaptability of the proposed parts feeder to
various types of parts and noise reduction during excitation have to be further
investigated in the future.
2.2.13
Simulation software for parts feeding in a vibratory bowl
feeder - 2003
This paper ([46]) describes the development of three-dimensional (3D) computer
simulation software to simulate parts feeding and orienting in a vibratory bowl
feeder. A mathematical model of part motion and the interaction of the part with
the orienting mechanism were developed. Based on the mathematical model, a
3D simulation software has been developed using Java. The key technology involves 3D object modelling and collision detection, which were developed with
54
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
Java 3D API. The simulation software can perform dynamic simulation of part
motion and the interaction of simple parts with the designed orienting mechanism, therefore enabling the reliability and capability of the design to be assessed prior to fabrication. The computer simulation was verified by carrying
out physical experiments in the laboratory. Good agreement between the results
of the computer simulation and experiments was obtained. The software can
perform simulation to evaluate the performance of part interaction with the orienting mechanism, which was designed by the user, and the designer can then
use the design information which is saved in the resource file by the software to
generate engineering drawings for fabrication.
2.2.14
Using air streams for part feeding systems - innovative
and reliable solutions for orientation and transport - 2003
This paper ([49]) describes the development of innovative feeding technologies
based on aerodynamic effects. The general idea of aerodynamic part feeding is
to create, e.g. a permanent air field which forces the part into the desired orientation without the need for any control by sensors. An exclusively aerodynamic
system offers the highest potential for achieving the goals of high process-speed
and inherently flexible feeding. The prototype implementation show how easily
and reliably such systems can be applied, although special aerodynamic knowledge is necessary. Further investigations of the IFA will be made to what kind of
workpieces and assembly systems are suitable for aerodynamic feeding.
2.2.15
Mechatronic design of a flexible vibratory feeding
system - 2003
This paper ([35]) describes the development of a sensor-based flexible vibratory
feeding system (SBFVFS) for agile manufacturing, which when fully developed
will consist of a decoupled vibratory feeder, a machine vision system, displacement sensors, power amplifiers and a computer system. Its major feeding parameters such as the vibration angle, frequency, acceleration amplitude and phase
difference can all be easily adjusted online by software means. Consequently,
the system can not only feed a wide range of parts without any retooling of the
feeder but can also feed various parts made of different materials in a way close
to the optimal state. The SBFVFS also has some built-in intelligence. It can find
out the natural frequencies of the decoupled vibratory feeder by means of an
automatic frequency response analysis and hence determine the best frequency
for the driving signals. It can also find out by machine inspection whether any
parts are jamming and eliminate them by making the parts move in reverse directions. A prototype of the SBFVFS has been successfully developed. Experimental
investigation is carried out based on the prototype and the results are promising.
2.2. TECHNIQUES FOR FLEXIBLE FEEDING
2.2.16
55
Chaotic dynamics of repeated impacts in vibratory bowl
feeders - 2002
The dynamic behavior of a single part on the vibrating track of the bowl feeder has been modelled and analyzed ([14]). The numerical simulation and experimental results for dynamic behavior and conveying rate in both periodic and
chaotic regimes are presented. The dynamic effects from the variation of several physical parameters are examined and the important features for the effective
design of the vibratory feeder are presented. While most of the previous studies were restricted to the purely periodic regime, the existence of chaotic regimes
is pointed out numerically and experimentally in this paper. The periodic and
chaotic region in hopping regime is identified through experiments and the feed
rates in each region were compared to numerical simulation results. It was verified experimentally that the conveying rate in the chaotic regime is more or less
independent of variations of the external parameters such as control parameter
and load. The results of experiments confirm the trends of numerical analysis for
the feed rate. Therefore, the simplified model and accompanying numerical analysis method can be used as an effective design tool for vibratory feeders. This
research holds much potential for leverage over design problems of a wide range
of mechanisms and tools with repeated collisions.
2.2.17
Generic flexible assembly system design - 2002
The development of a generic flexible assembly system involves the design, selection and integration of a number of different mechanical systems in order to
develop an assembly system, which is capable of assembling a wide variety of
products having an unknown specification. A specific system configuration being dependent on a variety of factors such as, product size, weight, component insertion direction, and manipulator geometry. This paper ([30]) examines each of
the factors that should be considered when designing a generic flexible assembly
system and presents a novel generic flexible assembly system design.
2.2.18
Orienting polyhedral parts by pushing - 2002
A common task in automated manufacturing processes is to orient parts prior
to assembly. The authors consider ([43]) sensorless orientation of an asymmetric polyhedral part by a sequence of push actions, and show that is it possible
to move any such part from an unknown initial orientation into a known final
orientation if these actions are performed by a jaw consisting of two orthogonal
planes. They also show how to compute an orienting sequence of push actions.
The authors propose a three-dimensional generalization of conveyor belts with
fences consisting of a sequence of tilted plates with curved tips; each of the plates
56
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
contains a sequence of fences. We show that it is possible to compute a set-up
of plates and fences for any given asymmetric polyhedral part such that the part
gets oriented on its descent along plates and fences.
2.2.19
Development of a model for part reorientation in
vibratory bowl feeders with active air jet tooling - 2001
Vibratory bowl feeders (VBFs) are widely used in industry for feeding and reorienting small parts in high volume production. This research ([22]) describes
the development of a model of part behavior required for reorienting a part with
an air-jet-based computer controlled orienting system. The control algorithm accepts the part’s weight, geometry, and its orientation. Sensors then compare the
present with the desired orientation and the algorithm determines the appropriate pulse of air to produce the desired orientation. The model was validated
experimentally for a specific part. This model can be used to reorient parts and
improve feed rates in sensor-based active air jet vibratory bowl feeder systems.
2.2.20
Flexible parts feeding for flexible assembly - 2001
This paper ([28]) describes a compact, low cost belt feeder prototype based on
Pherson et al.’s (1983) concept but, which is capable of feeding complex geometries using modern sensor technology for part recognition, a standard non-active
orientation blade, and a novel method for handling cylindrical parts.
Limitations
As with all part feeding systems, the belt feeder described in this paper has limitations; components need to have at least one non-rolling stable orientation; they
need to be physically robust enough to circulate in the feeder; they must not be
transparent. There are also component size limitations: with the current pusher
dimensions, components less than 12 mm in length and greater than 50 mm wide
are difficult to separate. Components less than 1 mm thick cannot be fed because
they jam under the belt guides. The feedrate is limited by the velocity of the conveyor belts and the likelihood of a correct part orientation.
Conclusions
• A flex-feeder based on the double belt feeder concept is potentially economic and versatile. It is possible to build a compact low-cost flex-feeder having an equivalent cost to a vibration bowl (£7000 to £9000). The prototype
flex-feeder was constructed for £5600.
• Vision system technology has developed enough to overcome the limitations of the early flex-feeders.
2.2. TECHNIQUES FOR FLEXIBLE FEEDING
57
• The double belt feeder described in this paper is more compact and cheaper
than other commercially available flexible feeding systems.
• The prototype flex-feeder would be an ideal replacement for a vibratory
bowl feeder where parts are subject to wide tolerance variation, e.g. distortion in plastic parts or there is a requirement to feed many diOEerent parts
in the same feeder.
• A low-cost pattern matching sensor can be used to identify part orientations, but is limited to parts that do not tangle or nest.
Further work
• A more advanced vision system will be used to simplify, and speed the
set-up operation between part changes.
• Three different sizes of belt feeder are proposed to cater for different component sizes.
• The effect of the step height between the return and feed belt will be studied.
2.2.21
New Feeding System for High Speed Assembly of Small
Parts - 2000
This paper ([39]) presents the new general purpose automatic feeding system,
developed for high speed assembly of small parts. The new design is a result of
the consequently applied Axiomatic Design Theory. The feeding system is functionally uncoupled, leading to minimization of problems in system tuning, parts
damage and noise emission. Moreover, the new design gives the possibility for
simple introduction of additional functional modules such as: sensorized modules for non-geometrical orientation of parts, or active orientation modules for
improving efficiency of the system. The proposed system is verified by the prototype developed for feeding screws on pneumatic screwdriving station. Future
research will address more strictly in the mathematical formulation of the proposed methodology, as well as the improvement of already developed feeding
system.
2.2.22
Designing a parts feeding system for maximum
flexibility - 1997
Short-run production, frequent product modifications, and pressures to reduce
product time to market make flexible manufacturing increasingly desirable. Despite advances in flexible systems, the problem of feeding parts to an assembly line is not fully addressed by conventional methods such as bowl feeders.
58
CAPITOLO 2. FLEXIBLE AUTOMATION STATE OF THE ART
New technologies ([16]) such as Intelligent Automation Systems’ FPF2000 Flexible Feeder for Small Parts offer additional versatility for short production cycles
with frequent changes, and multiple simultaneous assembly lines.
Capitolo 3
T HE ARTICLE SELECTED
I
we present the papers of J. Krüger, T.K. Lien, A. Verl (2009)
and M. Morioka, S. Sakakibara (2010).
N THIS CHAPTER
3.1
Cooperation of human and machines in
assembly lines
Flexibility and changeability of assembly processes require a close cooperation
between the worker and the automated assembly system. The interaction between human and robots improves the efficiency of individual complex assembly processes, particularly when a robot serves as an intelligent assistant. The
paper gives a survey about forms of human-machine cooperation in assembly
and available technologies that support the cooperation. Organizational and economic aspects of cooperative assembly including efficient component supply and
logistics are also discussed.
3.1.1
Introduction
Skilled robotic systems are key components in fully automated assembly processes as a part of a highly efficient production. Due to smaller lot sizes of customized products the demands for increased flexibility and adaptability to changing assembly tasks are rising continuously. Therefore a robot assisted but human
guided assembly has various significant advantages compared to full automation
(Bley H, Reinhhart G, Seliger G, Bernardi M, Korne T, 2005).
Flexibility and changeability of assembly processes require a close linkage between the worker and the automated assembly system. The interaction between
human and robot improves complex assembly processes, particularly when a
robot can be guided by a worker and the robot provides power assistance to the
worker.
59
60
CAPITOLO 3. THE ARTICLE SELECTED
Various forms of support can be provided to the worker in manufacturing
processes, depending on the degree of assistance with sensors, actuators or data
processing (Fig. 3.1). For assembly processes robot assistance combines these
components in order to give support for difficult, monotonous or exhausting
tasks.
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
Fig. 1. Influence factors on robot assistance [2].
Figura 3.1: Influence factors on robot assistance (www.assistor.de).
3.1.2
Human-machine cooperation in assembly lines - state of
the art
3.1.2.1
Hybrid assembly
Fig. 3. Cockpit install with IAD
The close linkage of human and machine in cooperative assembly tasks should
2.2. Workplace sharing systems
make use of the strengths of both sides. Typically an automated assembly
system
workplace
sharing systems robots and h
provides a couple ofFig.advantages
asanoperation
without
breaks Inand
fatigue
2. Combination of asuch
manual and
automatic work station
[3].
working in the same workplace, and both are
and high productivity for simple assembly tasks. Though, flexibility
ofasautomattasks
well as assembly tasks in two differ
engine
block
handling,
ed systems such as robots normally is restricted due to high programming effort
floor pan transfer,
Either the robot is performing an assembl
and limited abilities
for
handling
of complex or limp parts.
front
sub-frame
transfer,
worker is performing a handling task,
instrument panel load,
or the robot is performing a handling task a
On the other hand a human provides incomparable sensomotoric abilities for
strut handling,
is performing an assembly task.
complex handling
tasks, can
quickly
transmission
handling
and adapt to new process sequences but is retransmission sequencing.
The interaction of the robot and the huma
stricted in force and
precision. Cooperative work stations combine
the advanthe avoidance of collisions; the robot will
This an
IADautomated
technology was
originally
developed
tages of a human and
system
(Fig.
3.2). by the US between robot and human being is below a gi
Fig. 5 shows an example of a workplace sh
company Cobotics in 2003 and is based on fundamental research
One way of utilizing
robotic and human capabilities at their best
is obtained
done by Colgate and Peshkin [8]. Further research for so-called
the team@work system of Fraunhofer IPA
scenario
components have to be a
powerwhere
amplifying
assist is
devices
(IPAD) was done
by Fraunhofer
in assembly system
there
a sequential
division
of tasks.
Theseveral
simple
IPK [3] (Fig. 4).
metal part which is supplied by a conveyo
tasks suited for robots
found
upstream
in the
The complex
frequently
worker
has to grip the components out of bin
Hybridare
assembly
systems
can be divided
intoline.
two groups:
these
parts
towards
varied tasks that give the assembled products their individual features
are
per- the conveyor system (h
put the components on the sheet met
workplace sharing systems and
formed downstream
by human
been used
to has to screw these
Afterwards,
the robot
workplace
and time operators.
sharing systems.Such hybrid lines have
advantage in industry for more than two decades (Lien TK, Rasch FO 2001).
Today, especially for the assembly of heavy or bulky parts, weight compensators/balancers are used. Since these systems do not compensate for inertial
1. Influence factors
robot assistance [2]. IN ASSEMBLY LINES 61
3.1. COOPERATION OF Fig.
HUMAN
ANDonMACHINES
Fig. 3. Cockpit install wi
2.2. Workplace sharing systems
In workplace sharing systems robots
Fig. 2. Combination of a manual and an automatic work station [3].
Figura 3.2: Combination
of a manual and an automatic work station (Krüger
J, Bernhar,
working
in the same workplace, and bo
dt R, Surdilovic D, Seliger G 2006).
tasks as well as assembly tasks in two
engine block handling,
floor pan transfer,
Either the robot is performing an ass
front sub-frame transfer,
worker is performing a handling tas
forces, even small
mistakes
lead
to work-related injuries (lower back pain,
spine
instrument
panel
load,
or the robot
is performing a handling
strut handling,
is performing an assembly task.
injuries) (Zaeh MF, Prasch M, 2007). According to statistics of the Occupation transmission handling and
1
al Safety & Health
Department
(OSHA) of the US Department of Labour
, more of the robot and the
transmission
sequencing.
The interaction
the
avoidance
than 30% of European manufacturing workers are affected by lower back painof collisions; the robo
This IAD technology was originally developed by the US
between robot and human being is belo
which brings with
it enormous
social
economic
costs.research
In order toFig.improve
5 shows an example of a workp
company
Cobotics in 2003
and and
is based
on fundamental
by Colgate
and of
Peshkin
[8]. Further
researchassist
for so-called
team@work
this situation, adone
careful
design
socalled
intelligent
systemsthe
(IAS)
or in- system of Fraunhofe
scenario several components have to
power amplifying assist devices (IPAD) was done by Fraunhofer
telligent automation
devices
(IAD) and their operating procedures metal
is necessary
IPK [3] (Fig.
4).
part which is supplied by a con
has to to
grip the components out
Hybrid assembly
systems can
be dividedand
intohuman
two groups:
when physical collaboration
between
machines
workers worker
also have
these parts towards the conveyor syst
follow ergonomic
targets. Fig. 3.3 shows an IAD for cockpit assembly.
put the components on the sheet
workplace sharing systems and
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
629
Afterwards, the robot has to screw t
workplace and time sharing systems.
Fig. 1. Influence factors on robot assistance [2].
Fig. 4. IPAD – a reliable approach for advanced material handling [3].
Fig. 3. Cockpit install with IAD [7].
Figura 3.3: Cockpit install with IAD (www.stanleyassembly.com).
2.2. Workplace sharing systems
Industrial applications of human-machine cooperation are mainly to be found
In workplace sharing systems robots and human beings are both
Fig. 2. Combination of a manual and an automatic work station [3].
working
in the same workplace,
and both are (www.stanleyassembly.com)
performing handling
in automotive industry.
Stanley
Automation
protasks as well as assembly tasks in two different configurations:
engine block handling,
vides IADs, which support the worker in assembly tasks such as
floor pan transfer,
front sub-frame transfer,
instrument panel load,
strut handling,
transmission handling and
transmission sequencing.
1
www.osha.eu.int
This IAD technology was originally developed by the US
ompany Cobotics in 2003 and is based on fundamental research
one by Colgate and Peshkin [8]. Further research for so-called
ower amplifying assist devices (IPAD) was done by Fraunhofer
PK [3] (Fig. 4).
Either the robot is performing an assembly task and the human
worker is performing a handling task,
or the robot is performing a handling task and the human worker
is performing an assembly task.
The interaction of the robot and the human worker is limited to
the avoidance of collisions; the robot will stop if the distance
between robot and human being is below a given security distance.
Fig. 5 shows an example of a workplace sharing hybrid system:
the team@work system of Fraunhofer IPA and IPK [9]. In this
scenario several components have to be assembled on a sheet
metal part which is supplied by a conveyor system. The human
Fig. 1. Influence factors on robot assistance [2].
62
CAPITOLO 3. THE ARTICLE SELECTED
• axle sequencing,
• cardboard blank handling,
• catalytic converter test cell,
• engine block handling,
Fig. 2. Combination of a manual and an automatic work station [3].
• floor pan transfer,
engine•block
handling,
front
sub-frame transfer,
floor pan transfer,
• instrument
panel load,
front sub-frame
transfer,
instrument panel load,
• strut handling,
strut handling,
transmission handling and
• transmission handling,
transmission sequencing.
• transmission sequencing.
This IAD technology was originally developed by the US
company Cobotics in 2003 and is based on fundamental research
done by Colgate and Peshkin [8]. Further research for so-called
power amplifying assist devices (IPAD) was done by Fraunhofer
IPK [3] (Fig. 4).
Hybrid assembly systems can be divided into two groups:
Fig. 3. Cockpit install with IAD [7].
2.2. Workplace sharing systems
In workplace sharing systems robots and human beings are both
working in the same workplace, and both are performing handling
tasks as well as assembly tasks in two different configurations:
Either the robot is performing an assembly task and the human
worker is performing a handling task,
or the robot is performing a handling task and the human worker
is performing an assembly task.
The interaction of the robot and the human worker is limited to
the avoidance of collisions; the robot will stop if the distance
between robot and human being is below a given security distance.
Fig. 5 shows an example of a workplace sharing hybrid system:
the team@work system of Fraunhofer IPA and IPK [9]. In this
scenario several components have to be assembled on a sheet
metal part which is supplied by a conveyor system. The human
worker has to grip the components out of bins, in order to transport
these parts towards the conveyor system (handling time) and to
put the components on the sheet metal parts (assembly).
Afterwards, the robot has to screw these components into the
This IAD technology was originally developed by the US company Cobotics
in 2003 and is based on fundamental research done by Colgate and Peshkin
(1996). Further research for so-called power amplifying assist devices (IPAD)
workplace sharing systems and
was done
bysharing
Fraunhofer
IPK (Krüger J et al 2006) Fig. 3.4.
workplace
and time
systems.
Fig. 4. IPAD – a reliable approach for advanced material handling [3].
Figura 3.4: IPAD - a reliable approach for advanced material handling (Krüger J et al.
2006).
Hybrid assembly systems can be divided into two groups:
• workplace sharing systems,
• workplace and time sharing systems.
3.1.2.2
Workplace sharing systems
In workplace sharing systems robots and human beings are both working in the
same workplace, and both are performing handling tasks as well as assembly
tasks in two different configurations:
• Either the robot is performing an assembly task and the human worker is
performing a handling task,
• or the robot is performing a handling task and the human worker is performing an assembly task.
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 63
The interaction of the robot and the human worker is limited to the avoidance
of collisions; the robot will stop if the distance between robot and human being
is below a given security distance.
Fig. 3.5a shows an example of a workplace sharing hybrid system: the team@work
system of Fraunhofer IPA and IPK (Thiemermann S, Schraft RD 2003). In this
scenario several components have to be assembled on a sheet metal part which
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2
630
is supplied by a conveyor system. The human worker has to grip the components out of bins, in order to transport these parts towards the conveyor system
(handling time) and to put the components on the sheet metal parts (assembly).
Afterwards, the robot has to screw these components into the sheet metal part
(assembly). If the worker is too slow, the robot will wait until the worker has
finished his tasks.
The task distribution between robot and human worker in this scenario is as
follows (Fig. 3.5b).
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
630
Fig. 5. Example of a workplace sharing hybrid system (team@work [9]).
Fig. 7. Example of a
In this cooper
very precise pos
assembled with p
part AB to a seco
pick part A0 . In be
will be assembly
The task distr
scenario is as fol
TimeTime
distribution
between human
and robot
in workplace
Fig. 5. Example of a workplace
hybrid system (team@work [9]).Fig. 6.(b)
(a) sharing
Example.
distribution
between
human
andsharing
robot.systems.
In this scenari
tool to the huma
Figura 3.5: Workplace
flexible assembl
sheet sharing
metal part systems.
(assembly). If the worker is too slow, the robot
Fig. 7. Example of a workplace and time sharing hybrid system (PowerMate [10]).
sharing human–m
will wait until the worker has finished his tasks [9].
is on novel inte
The task distribution between robot and human worker in this
integration as we
scenario is as follows (Fig. 6).
In this cooperation mode the human operator is able to ensure a
equipment. The
very precise positioning of part A, so that part A can easily be
but to support
2.3. Workplace and time sharing systems
assembled with part B. As soon as this task is done, the robot moves
design of a sustai
part AB to a second bin and moves back to the first bin in order to
using human–ro
In workplace and
time sharing systems the human worker and
pick part A0 . In between the human operator prepares part B0 which
robot are additionally able to jointly perform
a handling task or an
0
will be assembly with part A in the next cycle.
assembly task at the same time, i.e. there are four different
3. Technologies
The task distribution between robot and human worker in this
configurations:
scenario is as follows (Fig. 8).
3.1. Interfaces be
Fig. 6. Time distribution between human and robot in workplace sharing systems.
In this scenario the robot serves like an intelligent and powerful
the robot is performing an assembly task and the human worker
tool to the human operator. The research project PISA deals with
is performing a handling task,
flexible assembly systems through workplace sharing and timeFor the effic
sheet metal part (assembly). If the worker is too slow, the robot
the robot is performing a handling task and the human worker is
machine-interfac
sharing human–machine cooperation [11]. The focus of the project
will wait until the worker has finished his tasks [9].
performing an assembly task,
assembly system
is on novel intelligent assist systems, planning tools for their
The task distribution between robot and human worker in this
the robot and the human worker are jointly performing a
integration as well as reconfigurability and reusability of assembly
scenario is as follows (Fig. 6).
handling task and
remote interfa
equipment. The overall goal is to keep human workers in the loop
the robot and the human worker are jointly performing an
gestures [12] a
but to support them with powerful tools. With respect to the
2.3. Workplace and time sharing systems
assembly task.
physical interf
design of a sustainable production several criteria can be improved
mounted displ
using human–robot interaction in assembly.
In workplace and time sharing systems the human worker and
robot are additionally able to jointly perform a handling task or anIn order to jointly handle or assemble objects the robot has to
interact 3.
with
the human worker
on a levelassembly
which is much higher
Robotic syste
assembly task at the same time, i.e. there are four different
Technologies
for cooperative
than
just
the
avoidance
of
collision.
graphical user i
configurations:
In the scenario which is shown in Fig. 7, the PowerMate system
3.1. Interfaces between human and machines
of Fraunhofer IPA [10], the interaction between robot and human
the robot is performing an assembly task and the human worker
worker is realized with a force-torque-sensor which enables that
is performing a handling task,
For the efficient cooperation and interaction the human–
the
can be moved by the human operator.
the robot is performing a handling task and the human worker
is robotmachine-interface
has an essential role. Interfaces for hybrid
The task of this hybrid system is to assemble heavy parts of an
performing an assembly task,
assembly systems can be divided into two classes:
automotive
rear axle. In order to do this, the robot has to grip part A
the robot and the human worker are jointly performing
a
out of a box and has to move it to the human worker. As these
handling task and
remote interfaces such as visual interfaces, interfaces for
take place in an area where the human operator has no
the robot and the human worker are jointly performingmovements
an
gestures [12] and voice and
access the robot is allowed to move at maximum speed. As soon as
assembly task.
physical interfaces such as haptic interfaces, displays and head
3.1.2.3
Workplace and time sharing systems
In workplace and time sharing systems the human worker and robot are additionally able to jointly perform a handling task or an assembly task at the same
time, i.e. there are four different configurations:
• the robot is performing an assembly task and the human worker is performing a handling task,
• the robot is performing a handling task and the human worker is performing an assembly task,
• the robot and the human worker are jointly performing a handling task,
• the robot and the human worker are jointly performing an assembly task.
64
CAPITOLO 3. THE ARTICLE SELECTED
In order to jointly handle or assemble objects the robot has to interact with
the human worker on a level which is much higher than just the avoidance of
collision.
Fig. 5. Example of a workplace sharing hybrid system (team@work [9]).
In the scenario which is shown in Fig. 3.6a, the PowerMate system of Fraunhofer IPA (Schraft RD, Meyer C, Parlitz Fig.
C,7.Helms
E, 2005), the interaction between
Example of a workplace and time sharing hybrid system (PowerMate [10]).
robot and human worker is realized with a force-torque-sensor which enables
that the robot can be moved by the human
operator.
In this
cooperation mode the human operator is able to ensure a
very precise positioning of part A, so that part A can easily be
The task of this hybrid system is toassembled
assemble
heavy parts of an automotive
with part B. As soon as this task is done, the robot moves
part AB
to a second
bin and
moves
the first
bin inand
order to
rear axle. In order to do this, the robot
has
to grip
part
A back
outto of
a box
pick part A0 . In between the human operator prepares part B0 which
has to move it to the human worker. As
movements
take
place
will these
be assembly
with part A0 in
the next
cycle.in an area
The task distribution between robot and human worker in this
where the human operator has no access
the robot is allowed to move at maxiscenario is as follows (Fig. 8).
Fig. 6. Time distribution between human and robot in workplace sharing systems.
this scenario
the robot servesthe
like an
intelligent
and powerful
mum speed. As soon as this part is in theIn shared
workspace
robot
stops
and
tool to the human operator. The research project PISA deals with
changes
to Ifa the
cooperation
mode.
In thisflexible
cooperation
mode
the human
worker
assembly systems
through
workplace sharing
andis
time
sheet metal part
(assembly).
worker is too slow,
the robot
sharing human–machine cooperation [11]. The focus of the project
will wait until the worker has finished his tasks [9].
able to move the robot by pushing, pulling or turning a special handling device
is on novel intelligent assist systems, planning tools for their
The task distribution between robot and human worker in this
integration
as well as reconfigurability
and reusability of assembly
scenario is aswhich
follows (Fig.
6).
is mounted
at the robot gripper in
combination
with a force-torque-sensor.
equipment. The overall goal is to keep human workers in the loop
this
cooperation
mode the human
is able
ensure
a very
precise
butoperator
to support them
withto
powerful
tools.
With respect
to the
2.3. Workplace andIn
time
sharing
systems
design
of
a
sustainable
production
several
criteria
can
improved
positioning of part A, so that part A can easily be assembled with part B. Asbesoon
using human–robot interaction in assembly.
In workplace and time sharing systems the human worker and
as thisable
task
is done,
the
robottask
moves
robot are additionally
to jointly
perform
a handling
or an part AB to a second bin and moves back to
assembly task at the same time, i.e. there are four different
the first bin in order to pick part A’ . 3.
InTechnologies
between for
thecooperative
humanassembly
operator prepares
configurations:
3.1. A’
Interfaces
between
and machines
part B’ which will be assembly with part
in the
nexthuman
cycle.
the robot is performing an assembly task and the human worker
task
human
inand
thisinteraction
scenario
as
is performing a The
handling
task,distribution between robot and
For the
efficientworker
cooperation
theishuman–
the robot is performing a handling task and the human worker is
machine-interface has an essential role. Interfaces for hybrid
follows
(Fig.
3.6b).
performingTechnology
an assembly
task,
assembly systems can be divided into two classes:
/ CIRP Annals - Manufacturing
58 (2009)
628–646
the robot and the human worker are jointly performing a
handling task and
the robot and the human worker are jointly performing an
assembly task.
remote interfaces such as visual interfaces, interfaces for
gestures [12] and voice and
physical interfaces such as haptic interfaces, displays and head
mounted displays (HMDs) as well as force feedback systems.
In order to jointly handle or assemble objects the robot has to
interact with the human worker on a level which is much higher
Robotic systems are typically operated via teach panels or
than just the avoidance of collision.
graphical user interfaces. An intuitive way of commanding a
In the scenario which is shown in Fig. 7, the PowerMate system
of Fraunhofer IPA [10], the interaction between robot and human
worker is realized with a force-torque-sensor which enables that
the robot can be moved by the human operator.
The task of this hybrid system is to assemble heavy parts of an
automotive rear axle. In order to do this, the robot has to grip part A
out of a box and has to move it to the human worker. As these
movements take place in an area where the human operator has no
access the robot is allowed to move at maximum speed. As soon as
this part is in the shared workspace the robot stops and changes to
eam@work [9]).
a cooperation mode. In this cooperation mode the human worker is
able to move the robot by pushing, pulling or turning a special
handling Fig.
device
which
is mounted
at sharing
the robot
gripper
in
7. Example
of a workplace
and
hybrid system
(PowerMate
[10]).
Fig.
8. Time
between human
and robothuman
in workplace
time sharing
(a)time
Example.
(b) distribution
Time distribution
between
andand
robot.
systems.
combination with a force-torque-sensor.
ace sharing systems.
slow, the robot
[9].
an worker in this
Figura 3.6: Workplace and time sharing hybrid system.
In this cooperation mode the human operator is able to ensure a
very precise positioning of part A, so that part A can easily be
assembled with part B. As soon as this task is done, the robot moves
part AB to a second bin and moves back to the first bin in order to
pick part A0 . In between the human operator prepares part B0 which
will be assembly with part A0 in the next cycle.
The task distribution between robot and human worker in this
scenario is as follows (Fig. 8).
In this scenario the robot serves like an intelligent and powerful
tool to the human operator. The research project PISA deals with
flexible assembly systems through workplace sharing and time
sharing human–machine cooperation [11]. The focus of the project
is on novel intelligent assist systems, planning tools for their
integration as well as reconfigurability and reusability of assembly
equipment. The overall goal is to keep human workers in the loop
but to support them with powerful tools. With respect to the
design of a sustainable production several criteria can be improved
In this scenario the robot serves like an intelligent and powerful tool to the
human operator. The research project PISA deals with flexible assembly systems
through workplace sharing and time sharing human-machine cooperation (Bernhardt R, Surdilovic D, Katschinski V, Schröer K,2007). The focus of the project is
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 65
on novel intelligent assist systems, planning tools for their integration as well as
reconfigurability and reusability of assembly equipment. The overall goal is to
keep human workers in the loop but to support them with powerful tools. With
respect to the design of a sustainable production several criteria can be improved
using human-robot interaction in assembly.
3.1.3
Technologies for cooperative assembly
3.1.3.1
Interfaces between human and machines
For the efficient cooperation and interaction the human-machine-interface has an
essential role. Interfaces for hybrid assembly systems can be divided into two
classes:
• remote interfaces such as visual interfaces, interfaces for gestures (Asfour
T, Berns K, Dillmann R, 2000) and voice and
• physical interfaces such as haptic interfaces, displays and head mounted
displays (HMDs) as well as force feedback systems.
Robotic systems are typically operated via teach panels or graphical user interfaces. An intuitive way of commanding a mobile robot assistant can be achieved
by verbal or gesture commands (Ehrenmann M, Lütticke T, Dillmann R, 2001,
Ehrenmann M, Becher R, Giesler B, et al, 2002, Theis C, Iossifidis I, Steinhage A,
2001). Due to environmental noise in assembly lines voice control of a robot may
be problematic whereas guidance by gestures may provide a robust form of communication between human and robot for short range remote control. The disadvantage of the use of voice or gesture is the single direction control providing
only visual feedback of robot motion. For many types of assembly tasks a closer
interaction between human and robot is needed, which includes a direct physical contact between both. Socalled admittance displays provide force feedback
enabling the operator to feel the contact between robot and assembly target by
rendering the contact impedance of the robot environment to the human (Krüger
J, Bernhardt R, Surdilovic D, Seliger G, 2006, Colgate JE, Wannasuphoprasit W,
Peshkin M, 1996).
3.1.3.2
Robotics
Assist robots
Helms et al. (2002) define a robot assistant as a direct interacting, flexible device,
that provides sensor based, actuator based and data processing assistance.
The assist robot rob@work was designed by the Fraunhofer Institute IPA. It is
a complex mechatronic system consisting of a mobile platform with differential
gear drive, energy supply for 9 h of work and a control system. Different tools
66
CAPITOLO 3. THE ARTICLE SELECTED
such as welding devices or drilling machines can be plugged to the seven DOF
manipulator. Target applications for rob@work are to be seen in small series and
lot size one manufacturing as well as in maintenance tasks (Fig. 3.7a).
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
631
631
Fig. 10. Robotic assist system [18].
3.2.2. Collaborative robots (COBOTS)
Collaborative robots or cobots invented by Edward Colgate
Fig. 9. Hardware architecture of rob@work [17].
[8,20] are(b)
mechanical
that[18].
provide guidance through the
Fig.
10.
Roboticdevices
assist system
(a) Hardware
architecture of rob@work (Schraft
Robotic
assist
system.
use of servomotors, while a human operator provides motive
RD, Helms E, Hans M, Thiemermann S, 2004).
power. Cobots
can
provide a virtual surface used to constrain and
3.2.2. Collaborative
robots
(COBOTS)
guide the
workers
mobile robot assistant can be achieved by verbal or gesture
Collaborative
robots
or motion.
cobots invented by Edward Colgate
Figura 3.7:lines
Assist
robot.
Fig. 9. Hardware
architecture
rob@work [17]. noise in assembly [8,20]
Another important
difference
of guidance
Cobots compared
commands
[13–15].
Due to of
environmental
are mechanical
devices that
provide
throughwith
the simple
balancers is the
ability
to provide
power support
to motive
the worker in a
voice control of a robot may be problematic whereas guidance
byof servomotors,
use
while
a human
operator
provides
way that
inertia
of surface
heavy work
can be reduced
by a
gestures may provide a robust form of communication between
power. Cobots
canapparent
provide a
virtual
usedpieces
to constrain
and
ten or
more, which means that physical strain of the worker
human
and robot
range
remote
control.
disadvantage
TM
guide thefactor
workers
motion.
mobile robot
assistant
canforbeshort
achieved
by verbal
or The
gesture
during
handling
of large
work pieces
is reduced
significantly
the use
of to
voice
or gesturenoise
is the
single direction
Another
important
difference
of Cobots
compared
with simple
commandsof
[13–15].
Due
environmental
in assembly
lines control
(Figs.
and 12).
providing
only
visual
feedback of robot
motion.
For many
of
balancers
is the11
ability
to provide power support to the worker in a
voice control
of a robot
may
be problematic
whereas
guidance
by types
For theinertia
support
the worker
during
handling
tasksby
with
assembly
tasks
a closer
interaction
between human
and robot
isthat apparent
way
of of
heavy
work pieces
can
be reduced
a heavy
gestures may
provide
a robust
form
of communication
between
loads
exoskeleton
systems
represent
an
alternative
approach.
needed,
which
includes
a
direct
physical
contact
between
both.
Sofactor ten or more, which means that physical strain of the worker Due to
human and robot for short range remote control. The disadvantage
the tight of
coupling
the robotic
axis the
force control of the
admittance
displays
provide
feedback
enablingduring
the
handling
large between
work pieces
is reduced
significantly
of the usecalled
of voice
or gesture
is the
singleforce
direction
control
robotic
operator
tofeedback
feel the contact
between
and assembly
by 11 and
(Figs.
12).system is complex [21]. Compared with collaborative robots
providing only
visual
of robot
motion.robot
For many
types of target
rendering
the interaction
contact impedance
the robot
environment
to the
For the support of the worker during handling tasks with heavy
assembly tasks
a closer
betweenofhuman
and
robot is
human
[3,8]. a direct physical contact between both. Soloads exoskeleton systems represent an alternative approach. Due to
needed, which
includes
The SMErobot
initiative (www.smerobot.org) offers an escape out of the
automation trap through:
• Technology development robot systems for small and medium enterprises
(SME) adaptable to varying degrees of automation, at a third of today’s
automation life-cycle costs,
the tight coupling between the robotic axis the force control of the
called admittance displays provide force feedback enabling the
3.2. Robotics
newbetween
business
models
options
forisfinancing
and operating
robot
robotic system
complex [21]. Compared
with collaborative
robots
operator to feel the•contact
robot and
assemblycreating
target by
rendering the contact impedance of the robot environment to the
given uncertainties in product volumes and life-times and to
3.2.1. Assist automation
robots
human [3,8].
Helms et al. [16] define a robot assistant as a direct interacting,
varying workforce qualification,
flexible device, that provides sensor based, actuator based and data
3.2. Robotics
processing
assistance.
• empowering the supply chain of robot automation by focusing on the needs
The assist robot rob@work was designed by the Fraunhofer
3.2.1. Assist robots
and
of SME manufacturing
Institute IPA.
It is culture
a complex mechatronic
system consisting of awith regard to planning, operation and
Helms et al. [16] define a robot assistant as a direct interacting,
mobile platform with differential gear drive, energy supply for 9 h
maintenance.
flexible device, that provides
sensor based, actuator based and data
of work and a control system. Different tools such as welding
processing assistance.
devices or drilling machines can be plugged to the seven DOF
The assist
robot rob@work
was designed
by
the Fraunhofer
manipulator.
Target applications
rob@work
are to be seen in
example
of afor
robot
assist
system
realized in the SMErobotT M initiative
Institute IPA.
It isseries
aAn
complex
consisting
a
small
and mechatronic
lot size one system
manufacturing
asof well
as in
mobile platform
with differential
gear
maintenance
tasks
(Fig.
9). drive, energy supply for 9 h
shows
Fig.
3.7b.
TM
of work and The
a control
system.
Different[18]
tools
suchanasescape
welding
SMErobot
initiative
offers
out of the
Fig. 11. Rail crane based cobot concept for assembly [3].
devices or automation
drilling Issofidis
machines
can
to the
seven DOF an anthropomorphic robot assistant for huet beal.plugged
(2002)
introduced
trap through:
manipulator. Target applications for rob@work are to be seen in
man
based on
asmall
seven
of freedom (DOF) manipulator arm in
small series
and lotenvironment
size
one manufacturing
as for
well
as and
in degree
Technology
development
robot systems
medium
maintenanceenterprises
tasks
(Fig.
9).
(SME) adaptable
degreesstereo
of automation,
at
combination
with toavarying
two DOF
camera
head. The robotic system comprises
TM
The SMErobot
initiative
[18] offers life-cycle
an escape
out of the
Fig. 11. Rail crane based cobot concept for assembly [3].
a third of
today’s automation
costs,
anthrough:
interface
to the
worker
forforthefinancing
interactive
correction of grasping orientation of
automation trap
new
business models
creating
options
and
operating robot automation given uncertainties in product
the
end
effector.
Technologyvolumes
development
robot systems
smallworkforce
and medium
and life-times
and to for
varying
qualification,
enterprises
(SME)
adaptable
to varying
degrees
of automation,
at focusing
empowering the supply
chain
of robot
automation by
a third of today’s
automation
life-cycle
costs,
on the needs
and culture
of SME
manufacturing with regard to
Collaborative
new business
models
creating
options (COBOTS)
for financing and
planning,
operation
androbots
maintenance.
operating robot automation given uncertainties in product
Collaborative
robots
or cobots
invented by Edward Colgate (Colgate JE et al.
example and
of a robot
assist workforce
system realized
in the SMErobotTM
volumes andAn
life-times
to varying
qualification,
initiative
shows chain
Fig. 10.
empowering
the supply
of robot automation by focusing
et al. of
introduced
an anthropomorphic
robot
on the needsIssofidis
and culture
SME manufacturing
with regard
to assistant
human environment
based on a seven degree of freedom (DOF)
planning,for
operation
and maintenance.
manipulator arm in combination with a two DOF stereo camera
headof
[19].
The robotic
systemrealized
comprises
toTM
the worker
An example
a robot
assist system
in an
theinterface
SMErobot
for theFig.
interactive
correction of grasping orientation of the end
Fig. 12. Rail crane based cobotic system providing power supply in order to reduce
initiative shows
10.
apparent inertia of heavy workpieces [23,24].
Issofidiseffector.
et al. introduced an anthropomorphic robot assistant
human and robot for short range remote control. The disadvantage
of the use of voice or gesture is the single direction control
providing only visual feedback of robot motion. For many types of
assembly tasks a closer interaction between human and robot is
needed, which includes a direct physical contact between both. Socalled admittance displays provide force feedback enabling the
operator to feel the contact between robot and assembly target by
rendering the contact impedance of the robot environment to the
human [3,8].
Fig. 10. Robotic assist system [18].
factor ten or more, which means that physical strain of the worker
during handling of large work pieces is reduced significantly
(Figs. 11 and 12).
For the support of the worker during handling tasks with heavy
loads exoskeleton systems represent an alternative approach. Due to
the tight coupling between the robotic axis the force control of the
robotic system is complex [21]. Compared with collaborative robots
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 67
3.2.2. Collaborative robots (COBOTS)
3.2. Robotics
Collaborative robots or cobots invented by Edward Colgate
[17].
[8,20]
are mechanical devices that provide guidance through the
3.2.1. Assist
robots
of [16]
servomotors,
while
a human
provides motive
Helmsuse
et al.
define a robot
assistant
as a operator
direct interacting,
power.that
Cobots
can provide
a virtual
surface
usedand
to constrain
and
flexible device,
provides
sensor based,
actuator
based
data
the workers motion.
erbal or gesture
processingguide
assistance.
Another
important difference
of Cobots
compared
with simple
in assembly lines
The assist
robot rob@work
was designed
by the
Fraunhofer
balancers
the ability
to provide power
support
to the
in a
reas guidance
by
Institute
IPA. It is aiscomplex
mechatronic
system
consisting
ofworker
a
way thatwith
apparent
inertiagear
of heavy
pieces
can for
be reduced
by a
nication between
mobile platform
differential
drive,work
energy
supply
9h
or more,
which
means that
physical
strain
of the worker
The disadvantage
of work factor
and a ten
control
system.
Different
tools
such as
welding
during
handling
of large
pieces
significantly
direction control
devices or
drilling
machines
can bework
plugged
to is
thereduced
seven DOF
(Figs.Target
11 andapplications
12).
or many types
of
manipulator.
for rob@work are to be seen in
the lot
support
the worker
during handling
with heavy
man and robot
is seriesForand
small
size ofone
manufacturing
as welltasks
as in
loadstasks
exoskeleton
etween both.maintenance
So(Fig. 9).systems represent an alternative approach. Due to
the tightTM
coupling
between
robotic
axis theout
force
of the
ack enabling the
The SMErobot
initiative
[18] the
offers
an escape
of control
the
Fig. 11. Rail crane based cobot concept for assembly [3].
robotic
is complex [21]. Compared with collaborative robots
sembly target
by
automation
trapsystem
through:
vironment to the
Technology development robot systems for small and medium
enterprises (SME) adaptable to varying degrees of automation, at
a third of today’s automation life-cycle costs,
new business models creating options for financing and
operating robot automation given uncertainties in product
direct interacting,
volumes and life-times and to varying workforce qualification,
or based and data
empowering the supply chain of robot automation by focusing
on the needs and culture of SME manufacturing with regard to
planning, operation and maintenance.
y the Fraunhofer
m consisting of a
gy supply for 9 h
An example of a robot assist system realized in the SMErobotTM
initiative shows Fig. 10.
such as welding
o the seven DOF
Issofidis et al. introduced an anthropomorphic robot assistant
forinhuman environment based on a seven degree of freedom (DOF)
are to be seen
arm in combination with a two DOF stereo camera
as well asmanipulator
in
head [19]. The robotic system comprises an interface to the worker
correction
of
grasping
offor
theassembly
end
cape out of for
thethe interactive
Fig. 12.
cranecrane
based cobotic
providing
powerproviding
supply in order
to reduce
Fig.
11.
Rail crane
basedcobot
cobotorientation
concept
for assembly
[3].
(a) Rail
crane
based
concept
(b)RailRail
basedsystem
cobotic
system
powapparent inertia of heavy workpieces [23,24].
effector.
mall and medium
of automation, at
financing and
nties in product
rce qualification,
ation by focusing
ng with regard to
n the SMErobotTM
c robot assistant
of freedom (DOF)
OF stereo camera
ace to the worker
ation of the end
1996, Akella M et al. 1998) are mechanical devices that provide guidance through
the use of servomotors, while a human operator provides motive power. Cobots
can provide a virtual surface used to constrain and guide the workers motion.
Another important difference of Cobots compared with simple balancers is
the ability to provide power support to the worker in a way that apparent inertia
of heavy work pieces can be reduced by a factor ten or more, which means that
physical strain of the worker during handling of large work pieces is reduced
significantly (Figs. 3.8a and 3.8b).
(Krüger J et al. 2006).
er supply in order to reduce apparent inertia
of heavy workpieces (Surdilovic D et al. 2005,
2007).
Figura 3.8: Cobots.
For the support of the worker during handling tasks with heavy loads exoskeleton systems represent an alternative approach. Due to the tight coupling
between the robotic axis the force control of the robotic system is complex (Kazerooni H, 1995). Compared with collaborative robots the exoskeleton systems provide a higher degree of mobility but on the other hand adaptation to the human
body is time-consuming.
Fig. 12. Rail crane based cobotic system providing power supply in order to reduce
apparent inertia of heavy workpieces [23,24].
Direct interaction between human and robot is also applied in surgery. In
these applications dynamic constraints are applied in order to control the authorized motions of the surgical tool held by the human operator during a planned
task (Delnondedieu Y, Trocazz J, 1995).
Robustness and stability of human-robot interaction
The most popular method of controlling physical human-robot interaction is to
vary the robots end point impedance by appropriate control strategies. Thereby
an implementation of haptic displays is possible by displaying varying impedances
/ admittances at the interaction point of human and robot.
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
632
the exoskeleton systems provide a higher degree of mobility but on
the other hand adaptation to the human body is time-consuming.
Direct interaction between human and robot is also applied in
surgery. In these applications dynamic constraints are applied in
order to control the authorized motions of the surgical tool held by
the human operator during a planned task [22].
68
CAPITOLO 3. THE ARTICLE SELECTED
Guaranteeing robust stability turned out to be a crucial challenge of such interactive systems. Therefore the coupled stability of interactive robots has been a
subject of intensive research in the past 25 years.
A variety of frameworks for coupled stability of interactive robots have been
developed but especially when interacting with unpredictable environments like
a human problems remain with these frameworks.
Fig. 14. Teaching the robot by manual guidance using force-torque-sensors [18].
Hogan and Colgate (1988) proposed a controller design framework using passivity of the controlled robot as a design criterion. For long years this was the
major method for designing interactive controllers and a variety of advanced
methods were developed on this foundation. One of those advanced method
is natural admittance control (NAC) (Dohring M, Newman W, 2003).
Hannaford (1998, 1999) derived a two-port framework to describe the interaction of users with virtual environments via haptic interfaces. Using this
framework he describes stable haptic interaction using passivity criteria.
For handling heavy objects comparatively stiff and heavy robots are needed.
The performance of a haptic interface to such a robot based on above mentioned
methods is very limited. Several research groups (Buerger SP, Hogan N, 2006,
Fig. 15. Humanoid portable robot for flexible assembly [36,37].
Sugano S, 2000) have derived methods of improving performance in high force
haptics using environment models and a robust
control
framework
humanoid
two-arm
manipulation (Colgate
robot whichetcan perform
complex tasks can work at a human’s workplace (Fig. 15).
3.2.4. Interactive learning of assist robot
al. 1989, Vukobratović M et al.2002, Book W et al.
1996).
Besides the humanoid construction the major difference of the
Fig. 13 shows a robot system, for which interactive control by
3.2.3. Robustness and stability of human–robot interaction
The most popular method of controlling physical human–robot
interaction is to vary the robots end point impedance by
appropriate control strategies. Thereby an implementation of
haptic displays is possible by displaying varying impedances/
admittances at the interaction point of human and robot.
Guaranteeing robust stability turned out to be a crucial
challenge of such interactive systems. Therefore the coupled
stability of interactive robots has been a subject of intensive
research in the past 25 years.
A variety of frameworks for coupled stability of interactive
robots have been developed but especially when interacting with
unpredictable environments like a human problems remain with
these frameworks.
Hogan and Colgate proposed a controller design framework
using passivity of the controlled robot as a design criterion [25,26].
For long years this was the major method for designing interactive
controllers and a variety of advanced methods were developed on
this foundation. One of those advanced method is natural
admittance control (NAC) [27].
Hannaford derived a two-port framework to describe the
interaction of users with virtual environments via haptic interfaces. Using this framework he describes stable haptic interaction
using passivity criteria [1,28,29].
For handling heavy objects comparatively stiff and heavy robots
are needed. The performance of a haptic interface to such a robot
based on above mentioned methods is very limited. Several
research groups [30,31] have derived methods of improving
performance in high force haptics using environment models and a
robust control framework [1,29,31–34].
the worker is realized through pointing to objects with laser
pointers. Within the SME Robot Project [18] technology for
simplified teaching the robot by guidance was developed based on
force-torque-sensors (Fig. 14).
robot arm shown in Fig. 16 compared to industrial robots is its light
weight structure providing intrinsic safety.
The Deutsches Zentrum für Luft- und Raumfahrt (DLR) designed
humanoid robots for application areas which are generally not
covered by industrial robots, but are still ongoing research topics.
Typical examples are:
Interactive learning of assist robot
Fig. 3.9a shows a robot system, for which interactive control by the worker is
3.2.5.
Humanoid/anthropomorphic
robotic systems
realized
through pointing
to objects with laser pointers. Within the SME Robot
A future flexible assembly line will incorporate different
Project
(www.smerobot.org)
technology
teaching
the robot
by guidworkplace
types
varying from fully automated
work cells for
to simplified
Assembly processes
for which
the position
estimation for the
manual
assembly
places.
Between
these
extremes
different
types
mating
parts
and/or
the
positioning
accuracy
of the robot is
ance was developed based on force-torque-sensors (Fig. 3.9b).
632
of cooperation and interaction between human and robot are
applied to achieve maximum flexibility. For this purpose a
significantly below the assembly tolerance,
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
the exoskeleton systems provide a higher degree of mobility but on
the other hand adaptation to the human body is time-consuming.
Direct interaction between human and robot is also applied in
surgery. In these applications dynamic constraints are applied in
order to control the authorized motions of the surgical tool held by
the human operator during a planned task [22].
3.2.3. Robustness and stability of human–robot interaction
The most popular method of controlling physical human–robot
interaction is to vary the robots end point impedance by
appropriate control strategies. Thereby an implementation of
haptic displays is possible by displaying varying impedances/
admittances at the interaction point of human and robot.
Guaranteeing robust stability turned out to be a crucial
challenge of such interactive systems. Therefore the coupled
stability of interactive robots has been a subject of intensive
Fig. 13.
worker
is pointingis
outpointing
objects directly
in theobjects
scene using
a laser pointer
Fig. 14.
the robot
manualby
guidance
usingguidance
force-torque-sensors
(a)TheThe
worker
out
directly
(b)Teaching
Teaching
thebyrobot
manual
using [18].
research in
the past 25 years.
Fig. 16. Humanoid light weight robot structure [38].
[35].
in the scene
laser pointer
(Stopp A,
force-torque-sensors.
A variety of frameworks
for using
coupleda stability
of interactive
robots have been Horstmann
developed butS,especially
when
Kristensen
S, interacting
Lohnert F,with
2003).
unpredictable environments like a human problems remain with
these frameworks.
Figura 3.9: Robot system.
Hogan and Colgate proposed a controller design framework
using passivity of the controlled robot as a design criterion [25,26].
For long years this was the major method for designing interactive
controllers and a variety of advanced methods were developed on
this foundation. One of those advanced method is natural
admittance control (NAC) [27].
Hannaford derived a two-port framework to describe the
interaction of users with virtual environments via haptic interfaces. Using this framework he describes stable haptic interaction
Hannaford derived a two-port framework to describe the
nteraction of users with virtual environments via haptic interaces. Using this framework he describes stable haptic interaction
sing passivity criteria [1,28,29].
For handling heavy objects comparatively stiff and heavy robots
re needed. The performance of a haptic interface to such a robot
P Annals - Manufacturing Technology 58 (2009) 628–646
ased on above mentioned
methods is very limited.
Several AND MACHINES IN ASSEMBLY LINES 69
3.1. COOPERATION
OF HUMAN
esearch groups [30,31] have derived methods of improving
Fig. 15. Humanoid portable robot for flexible assembly [36,37].
obility but on
erformance in high force haptics using environment models and a
-consuming.
obust control framework [1,29,31–34].
so applied in
humanoid two-arm manipulation robot which can perform
re applied in
complex tasks can work at a human’s workplace (Fig. 15).
.2.4. Interactive learning of assist robot
l tool held by
Humanoid/anthropomorphic
robotic
systems
Besides the humanoid construction the major difference of the
Fig. 13 shows a robot
system, for which interactive control
by
robot
arm different
shown in Fig.
16 compared
to industrial
robots is its light
he worker is realized
through
pointing
to
objects
with
laser
A future flexible assembly line will incorporate
workplace
types
varying
weight structure providing intrinsic safety.
ointers. Within the SME Robot Project [18] technology for
tion
from
automated
work based
cells on
to manual
assemblyZentrum
places.
these exThe Deutsches
fürBetween
Luft- und Raumfahrt
(DLR) designed
implified teaching the
robotfully
by guidance
was developed
uman–robot
humanoid
robots for
application
areas and
which
are generally not
orce-torque-sensors (Fig.
14). different types of cooperation and
tremes
interaction
between
human
robot
pedance by
covered by industrial robots, but are still ongoing research topics.
are applied
to achieve maximum flexibility.
For this purpose a humanoid twomentation
of
Typical examples are:
.2.5. Humanoid/anthropomorphic
robotic systems
impedances/
manipulation
robot which
can perform complex tasks can work at a human’s
A future flexible arm
assembly
line will incorporate
different
obot.
workplace
types varying
from
fully
automated
work
cells
to
Assembly processes for which the position estimation for the
workplace (Fig. 3.10a).
be a crucial
manual
assembly places. Between these extremes different types
mating parts and/or the positioning accuracy of the robot is
the
coupled and interaction between human and robot are
f cooperation
significantly below the assembly tolerance,
of intensive
pplied
to achieve
maximum
flexibility.
For
this
purpose
a
Fig. 14. Teaching the robot by manual guidance using force-torque-sensors [18].
f interactive
eracting with
remain with
n framework
erion [25,26].
g interactive
developed on
d is natural
describe the
haptic interc interaction
heavy robots
such a robot
ted.
Several
ig.
13. The
worker is pointing out objects directly in the scene using a laser pointer
f improving
35].
(a) Humanoid
portable
robotrobot
for flexible
assembly
Fig. 15. Humanoid
portable
for flexible
assembly(Bernhardt
[36,37].
models and a
R et al. 2008).
ve control by
s with laser
hnology for
ped based on
ate different
ork cells to
fferent types
nd robot are
s purpose a
Fig.
16.Humanoid
Humanoid light
robot
structure
[38].
(b)
lightweight
weight
robot
struc-
ture (Albu-Schäffer A et al. 2007).
Figura 3.10:robot
Humanoid/anthropomorphic
robotic.
humanoid two-arm manipulation
which can perform
complex tasks can work at a human’s workplace (Fig. 15).
Besides the humanoid construction the major difference of the
robot arm shown in Fig. 16 compared to industrial robots is its light
Besides the humanoid construction the major difference of the robot arm shown
weight structure providing intrinsic safety.
in
3.10bZentrum
compared
to industrial
robots
is its light weight structure providing
TheFig.
Deutsches
für Luftund Raumfahrt
(DLR) designed
humanoid
robots
for
application
areas
which
are
generally
not
intrinsic safety.
covered by industrial robots, but are still ongoing research topics.
Theexamples
Deutsches
Typical
are: Zentrum für Luft-und Raumfahrt (DLR) designed humanoid
robots for application areas which are generally not covered by industrial robots,
Assembly processes for which the position estimation for the
but are still ongoing research topics.
mating parts and/or the positioning accuracy of the robot is
significantly
below
the assembly
Typical
examples
are: tolerance,
• Assembly processes for which the position estimation for the mating parts
and/or the positioning accuracy of the robot is significantly below the assembly tolerance,
• applications in which the robot works in immediate vicinity of humans and
possibly in direct physical cooperation with them,
applications in which the robot works in immediate vicinity of
humans and possibly in direct physical cooperation with them,
mobile service robotics applications (arms mounted on mobile
platforms), for which the information about the position of the
robot and the surrounding objects, as well as about the
dimension of these objects is afflicted with relatively high
uncertainty [38].
70
CAPITOLO 3. THE ARTICLE SELECTED
•
3.2.6. Portable robots
The workspace of portable assist systems is not restricted to the
mobile service
robotics
(armswithin
mounted
on mobile platforms),
place of
its actual applications
use but can be chosen
the production
environment.
This
innovation
comprises
a
couple
of
advantages
for which the information about the position of the robot for
and the surroundmanufacturing and assembly:
ing objects, as well as about the dimension of these objects is afflicted with
Due to
a parallel task(Albu-Schäffer
operation of man and
efficiency
relatively high
uncertainty
A machine,
et al. 2007).
can be increased,
cost reduction and improvement of ergonomics is achieved by
use of the specific strengths of human and machine,
flexibility and adaptability with respect to place of installation of
handling technology, capacity, experience and knowledge of staff
robots and also with respect to type and complexity of the assembly
task can be improved [2].
Fig. 18. The mechatronic joint design of the LWR
and sensing [38].
Portable
The workspace of portable assist systems is not restricted to the place
itsforacthan of
15 kg
arms with a work space
dynamic
performance. There should b
order to within
improve the
the flexibility
of industrial
robots for This innovation
tual use but can beIn chosen
production
environment.
robot and no electronics cabinet as u
material handling, a portable robot system has been developed by
comprises a couple
of advantages for manufacturing and assembly: industrial robots.
Brecher et al. [39] within the project PORTHOS.
•
•
The full state measurement in all joi
The robot system can be installed at machine tools for
cycle, using
feeding of workpieces (Fig. 17). For the generation of robot
Due to a programs
parallel atask
operation of man and machine, efficiency can be
programming system is introduced that combines
strain gauge-based torque-sensing,
increased,the traditional approaches of online- and offline-programming
motor position sensing based on m
in order to allow fast and easy reprogramming of material
cost reduction
and improvement of ergonomics is achieved byand
use of the
handling tasks. The main feature of the programming system is
link-side position sensing based on po
specific strengths
ofuser-interface
human and
machine,
an intuitive
that
can be operated without special
redundant sensors for safety conside
qualifications.
• flexibility and adaptability with respect to place of installation of handling
3.4. Safety
technology,
and knowledge of staff and also
withsystems
re3.3.capacity,
Sensors andexperience
actuators
spect to typeBesides
and complexity
of the assembly task can be improvedThe
(www.assistor.de).
desired coexistence of robotic s
advanced control strategies to achieve a compliant and
same physical domain, by sharing t
thereby safe robot behaviour a second approach favours the use of
cooperating in a physical manner, po
inherently compliant
actuatorsof
to guarantee
safety
during human–
In order to improve
the flexibility
industrial
robots
for material
handling,
problem of ensuring safety for the user
robot interaction [40,41].
a portable robot system
been
by Brecher
al.18)
(2005) within
the norm ‘‘DIN EN 775
The European
The mainhas
design
goalsdeveloped
of the DLR lightweight
robotset
(Fig.
robots’’ which is right now revised
were
to
build
a
manipulator
with
kinematic
redundancy
similar
to
project PORTHOS.
international standard ‘‘ISO 10218
the human arm, i.e. with seven degrees of freedom (DOF), a load– Safety requirements’
The robot system
canratio
be of
installed
at machine
tools
for feeding
workpieces
to-weight
approximately
1:1 where
industrial
robots ofenvironments
for safety in a robot cell. Already with
typically
have
a
ratio
of
1:10
or
lower,
a
total
system
weight
of
less
(Fig. 3.11).
DIN EN 775 the possible installation
reach of a human was considered
conditions. One of the targets of th
further provide regulations for the r
[42]. Fig. 19 shows the safety-related
cells to human–robot interaction.
Fig. 19. Safety control – from the classical robot
17. Setup of the PORTHOS robot system [39].
[42].
Figura 3.11: Setup Fig.
of the
PORTHOS robot system (Brecher et al. 2005).
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 71
For the generation of robot programs a programming system is introduced
that combines the traditional approaches of online-and offline-programming in
order to allow fast and easy reprogramming of material handling tasks. The
main feature of the programming system is an intuitive user-interface that can
be operated without special qualifications.
3.1.3.3
Sensors and actuators
Besides advanced control strategies to achieve a compliant and thereby safe robot
behaviour a second approach favours the use of inherently compliant actuators
to guarantee safety during human-robot interaction (Som F, 2006, Zinn M, Khatib
O, Roth B, 2004).
The main design goals of the DLR lightweight robots (Fig. 3.12) were to
build a manipulator with kinematic redundancy similar to the human arm, i.e.
with seven degrees of freedom (DOF), a load-to-weight ratio of approximately
1:1 where industrial robots typically have a ratio of 1:10 or lower, a total system
weight of less than 15 kg for arms with a work space of up to 1.5 m, and a high
dynamic performance. There should be no bulky wiring on the robot and no
electronics cabinet as usually required by typical industrial robots.
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
633
robot works in immediate vicinity of
direct physical cooperation with them,
pplications (arms mounted on mobile
information about the position of the
ding objects, as well as about the
ects is afflicted with relatively high
le assist systems is not restricted to the
can be chosen within the production
on comprises a couple of advantages for
ly:
ration of man and machine, efficiency
vement of ergonomics is achieved by
ths of human and machine,
y with respect to place Figura
of installation
3.12: ofThe
city, experience and knowledge of staff
and
type and complexity of the assembly
Fig. 18. The mechatronic joint design of the LWR including actuation, electronics,
mechatronic
and sensing [38]. joint design of the LWR including actuation, electronics,
sensing (Albu-Schäffer A et al. 2007).
than 15 kg for arms with a work space of up to 1.5 m, and a high
performance.
should
no bulky wiring
the cycle, using
he flexibility of industrialThe
robots
fullforstatedynamic
measurement
in There
all joints
isbe
performed
in a on
3 kHz
robot and no electronics cabinet as usually required by typical
e robot system has been developed by
industrial robots.
he project PORTHOS.
The full state
measurement in all joints is performed in a 3 kHz
be installed at machine
• tools
strainforgauge-based
torque-sensing,
cycle, using
g. 17). For the generation of robot
• motor position sensing based on magneto-resistive encoders, and
system is introduced that combines
strain gauge-based
torque-sensing,
of online- and offline-programming
• link-side position
sensing based
on potentiometers (used only as redundant
motor position sensing based on magneto-resistive encoders,
nd easy reprogramming of material
sensors for safety
considerations).
and
eature of the programming system is
link-side position sensing based on potentiometers (used only as
that can be operated without special
redundant sensors for safety considerations).
l strategies to achieve a compliant and
ur a second approach favours the use of
3.4. Safety systems
The desired coexistence of robotic systems and humans in the
same physical domain, by sharing the same workspace and
without special
link-side position sensing based on potentiometers (used only as
redundant sensors for safety considerations).
3.4. Safety systems
72
CAPITOLO
3. THE
The desired coexistence of robotic
systems
and ARTICLE
humans inSELECTED
the
a compliant and
same
physical
domain,
by
sharing
the
same
workspace
and
avours the use of
in a physical manner, poses the very fundamental
Safety systems
y during human–3.1.3.4cooperating
problem of ensuring safety for the user and the robot.
The desired
robotic
systems
same physical doThecoexistence
European ofnorm
‘‘DIN
EN and
775 humans
Safety in
of the
manipulating
t robots (Fig. 18)
main, robots’’
by sharing
the same
workspace
and cooperating
in a physical
manner,
which
is right
now revised
and converted
into the
ndancy similar to
the very fundamental
problem
for the
and the robot.
standard
‘‘ISOof ensuring
10218: safety
Robots
for user
industrial
m (DOF), a load-poses international
European norm
“DIN EN
775 Safety offorms
manipulating
robots”
which is
environments
– Safety
requirements’’
the current
basis
ndustrial robots The
nowsafety
revised
converted
into the
standard of
“ISO
in aand
robot
cell. Already
withinternational
the implementation
the 10218:
em weight of lessright for
Robots
for industrial
environments
- Safety requirements”
thewithin
current basis
DIN
EN 775 the
possible installation
of a robot forms
system
reach
a human
was with
considered
under highest
security
for safety
in aof
robot
cell. Already
the implementation
of the DIN
EN 775 the
conditions.
One
the system
targetswithin
of the
ISOof 10218
now
to
possible
installation
of aofrobot
reach
a human
wasisconsidered
further
provide
regulations
for
the
robot–human-cooperation
under highest security conditions. One of the targets of the ISO 10218 now is to
[42].
Fig. 19
shows the
steps from classical
robot
further
provide
regulations
forsafety-related
the robot-human-cooperation
(Oberer
S, Schraft
cells
to
human–robot
interaction.
R-D, 2007). Fig. 3.13 shows the safety-related steps from classical robot cells to
[39].
human-robot interaction.
Fig.3.13:
19. Safety
– from
the
classical
robot
cellcell
to human–robot
interaction
Figura
Safetycontrol
control
- from
the
classical
robot
to human-robot
interaction.
[42].
Pre-collision systems
Control based limitations and sensor based surveillance of the workspace
REIS-Robotics introduced a control feature for their robot systems, with which
flexibly programmable limitations for automatic operation can be defined. The
robot arm exclusively moves within these restricted areas (Fig. 3.14). These regions can be programmed by direct teach-in, which means in a haptic way by use
of a 6-DOF-sensor at the end effector, or in a classical way by teach-in panel or
offline-programming/robot simulation system.
The changing characteristics of robot processes, with increasing payloads,
work ranges and cycle times necessitate a more flexible approach to safety, which
Fig. 20.
Safety controller
for robot
assisted welding [43].
3.1. COOPERATION OF
HUMAN
AND
MACHINES
IN ASSEMBLY LINES 73
634
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
3.4.1. Pre-collision systems – control based limitations and sensor
based surveillance of the workspace
REIS-Robotics introduced a control feature for their robot
systems, with which flexibly programmable limitations for
automatic operation can be defined. The robot arm exclusively
moves within these restricted areas (Fig. 20). These regions can be
programmed by direct teach-in, which means in a haptic way by
Fig. 22. SafetyEYE camera for work
use of a 6-DOF-sensor at the end effector, or in a classical way by
teach-in panel or offline-programming/robot simulation system.
to the machine controller. If a detectio
The changing characteristics of robot processes, with increasing
configurable outputs are shut down. T
payloads, work ranges and cycle times necessitate a more flexible
is located above the workstation for
approach to safety, which cannot be addressed with traditional
operating range (Fig. 22 lower part).
methods. Conventional safety relay technology has also restricted
immediate danger zone, there will b
functionality of safety systems, particularly in terms of flexibility
Fig. 20. Safety controller for robot assisted welding [43].
Figura 3.14:
Safety
controller
for robot
assisted
welding a(Kroth
work. When a worker enters the zone
and diagnostics.
Kuka Roboter
GmbH
has developed
safety E, 2007).
several seconds to reach, the contro
system for industrial robots incorporating the safety-related
When the worker steps back, the rob
fieldbus,
SafetyBUS
p,
in
cooperation
with
Pilz
GmbH.
The
3.4.1. Pre-collision systems – control based limitations and sensor
For safe human–robot interaction
Electronic
Safety Circuit
(ESC), coupled with SafetyBUS p and Pilz
based surveillance
of the workspace
cannot be addressed
with
traditional
methods.
Conventional
safety
relay
tech-the robot control ar
tion
concepts
PSSREIS-Robotics
safety
controllers,
is
now
being
used
by
BMW
at
its
Body-inintroduced a control feature for their robot
component
ensure safe operation of
White
line
in
Dingolfing,
Germany
[44].
withfunctionality
which flexibly ofprogrammable
limitations
for
nology has alsosystems,
restricted
safety systems,
particularly
in termstoof
robot controllers are developed whic
Standardoperation
automated
protection
devices
use laser
automatic
can optical
be defined.
The robot
arm exclusively
flexibility and diagnostics.
Kuka
Roboter
GmbH
has
developed
a safety
system
resident
in robots workspace during o
scanner
technology
in order
to separate
human
from
moves within
these restricted
areas
(Fig. 20).the
These
regions
canthe
be
control
was
by Kuka follow
robot.
The
major
supplier
for
these
type
of
safety
systems
for
for industrial robots
incorporating
the which
safety-related
fieldbus,
in
programmed
by direct teach-in,
means in a haptic
way bySafetyBUS p, developed
Fig.
22. Kuka’s
SafetyEYE
camera for
work
and
Fanuc
[47].
solution
is stric
automated
assembly
is
the
German
company
SICK
[45].
Fig.
21
use of a 6-DOF-sensor at the end effector, or in a classical way by
cooperation with
Pilz GmbH.use
The
Electronic
Safety Circuit
(ESC), coupled
with improved system re
on the severely
demonstrates
of SICK
active opto-electronic
protection
teach-in panelthe
or offline-programming/robot
simulation
system.
handing
over
safety
responsibilities
di
devices
(AOPD)
assembly.
SafetyBUS p and
Pilz
PSS in
safety
controllers,
is now with
being
used by to
BMW
at its
the machine
controller.
If a detection
The
changing
characteristics
of robot processes,
increasing
The
solution
by
Reis
in
contrast
rel
SafetyEYE
by
Pilz
is
a
camera
System
for
3D
workplace
configurable outputs are shut down. T
payloads,
work ranges and
cycle times necessitate a more flexible
Body-in-Whitesurveillance
line in Dingolfing,
controller
robot operation
22).
It Germany.
is cannot
meant be
to addressed
erase the with
need traditional
for safety
is
located monitoring
above the workstation
for
approach to (Fig.
safety,
which
control architecture.
different
co
fences
separating
the robot
workspace
from thehas
worker
[46]. The
(www.industrialnetworking.co.uk/mag/v9-1/f_safety.html)
operating
range (Fig. The
22 lower
part).
methods.
Conventional
safety
relay technology
also restricted
enter the robot’s
workspace
while
th
system
has a of
sensing
device, a particularly
computer, and
a programmable
immediate
danger
zone,
there
will
be
functionality
safety
systems,
in
terms
of
flexibility
Standard automated
optical
protection
devices has
usethree
laser
scannerslowing
technology
robot
operation
to athe
safe
mo
safety
and controlKuka
system.
The sensing
cameras
work. When
a worker
enters
zone
and diagnostics.
Roboter
GmbH device
has developed
a
safety
in order to separate
the
human
from
the
robot.
The
major
supplier
for
these
For
close
forms
of
cooperation,
it is
(Fig.
22
upper
part).
The
computer
receives
the
camera’s
image
several seconds to reach, the contro
system for industrial robots incorporating the safety-related
motionthe
in 3D
but also
toback,
have the
an exac
data
via fiber-optic
cables
and
creates a three-dimensional
image
When
worker
steps
rob
fieldbus,
SafetyBUS
p,
in
cooperation
with
Pilz
GmbH.
The
type of safety systems for automated assembly is the German company SICK
body.
stereoscopic
approach
for 3D
using
complex
algorithms.
Thiscoupled
image iswith
thenSafetyBUS
superimposed
ForAsafe
human–robot
interaction
Electronic
Safety
Circuit (ESC),
p andover
Pilz
(www.sick.de).the
Fig.
3.15
demonstrates
the
use
of
SICK
active
opto-electronic
was
developed
by
Fraunhofer
IPK
ac
detection
zones
image
to
see
any
zone
violation.
The
computer
tion concepts the robot controlinarc
PSS safety controllers, is now being used by BMW at its Body-inIPA
within
the
project
team@work
gives
results
to
the
safety
system
controller,
which
is
the
interface
component
to
ensure
safe
operation
of
White
line
in
Dingolfing,
Germany
[44].
protection devices (AOPD) in assembly.
detectcontrollers
segments ofare
skindeveloped
within thewhic
ima
robot
Standard automated optical protection devices use laser
characteristic
points
of the body.
Base
resident
in robots
workspace
during
o
scanner technology in order to separate the human from the
photogrammetric
3D
capturing
of
the
control was developed by Kuka follow
robot. The major supplier for these type of safety systems for
system
that
enables
clos
andAnother
Fanuc [47].
Kuka’s
solution
is stric
automated assembly is the German company SICK [45]. Fig. 21
PMD
flexible
work
on
thecamera
severelybased
improved
system
re
demonstrates the use of SICK active opto-electronic protection
developed
bysafety
Fraunhofer
IPA. Thedir
c
handing
over
responsibilities
devices (AOPD) in assembly.
use
the
cameras
distance
data
and
the
The solution by Reis in contrast rel
SafetyEYE by Pilz is a camera System for 3D workplace
and to generate
a signal
as output
wh
controller
monitoring
robot
operation
surveillance (Fig. 22). It is meant to erase the need for safety
risk, warn
and stop [49].
control
architecture.
The different co
fences separating the robot workspace from the worker [46]. The
Thethe
acquired
datawhile
fromth
th
enter
robot’sdistance
workspace
system has a sensing device, a computer, and a programmable
processed,
the
foreground
extracted
slowing robot operation to a safe mo
safety and control system. The sensing device has three cameras
Cartesian
space
to get
the surface of
For close
forms
of cooperation,
it th
is
(Fig. 22 upper part). The computer receives the camera’s image
workspace.
the an
robot
motion
in 3DAdditionally,
but also to have
exaci
data via fiber-optic cables and creates a three-dimensional image
Cartesian
space. By comparing
both
body.
A stereoscopic
approach for
3Do
using complex algorithms. This image is then superimposed over
model
and
objects
seen
from
the
was developed by Fraunhofer IPK in a c
the detection zones image to see any zone violation. The computer
identified
the camera
object[
IPA
withininside
the project
team@work
gives results to the safety system controller, which is the interface
treatedsegments
as obstacle
and evaluated
detect
of skin
within the agai
ima
21. Application of active opto-electronic protection devices in assembly [45].
the robot [49].points of the body. Base
characteristic
Figura 3.15: Fig.
Application
of active opto-electronic protection devices in assembly.
photogrammetric 3D capturing of the
Another system that enables clos
PMD camera based flexible works
SafetyEYE by Pilz is a camera System for 3D workplace surveillance
developed(Fig.
by Fraunhofer IPA. The co
useworkspace
the cameras distance data and the
3.16). It is meant to erase the need for safety fences separating the robot
and to generate a signal as output wh
from the worker (www.pilz.de/products/sensors/camera/f/safetyeye/index.jsp).
risk, warn and stop [49].
The acquired distance data from th
processed, the foreground extracted
Cartesian space to get the surface of th
workspace. Additionally, the robot is
Cartesian space. By comparing both o
model and objects seen from the
74
CAPITOLO 3. THE ARTICLE SELECTED
The system has a sensing device, a computer, and a programmable safety and
control system. The sensing device has three cameras (Fig. 3.16 left part). The
computer receives the camera’s image data via fiber-optic cables and creates a
three-dimensional image using complex algorithms. This image is then superimposed over the detection zones image to see any zone violation. The computer
gives results to the safety system controller, which is the interface to the machine
controller. If a detection zone has been violated, the configurable outputs are
shut down. Typically, the sensing device is located above the workstation for an
overview of the robot’s operating range (Fig. 3.16 right part). When a worker
enters the immediate danger zone, there will be an emergency stop to the work.
When a worker enters the zone where the robot would take several seconds to
reach, the controls reduce the robot speed. When the worker steps back, the robot
returns to normal speed.
Figura 3.16: SafetyEYE camera for workplace surveillance.
For safe human-robot interaction besides sensor and interaction concepts the
robot control architecture is an important component to ensure safe operation of
the robot. Therefore failsafe robot controllers are developed which enable the
worker to be resident in robots workspace during operation. The first safe robot
control was developed by Kuka followed by solutions by Reis, ABB and Fanuc
(Kochan A, 2006). Kuka’s solution is strictly software based and relies on the
severely improved system response time to failures by handing over safety responsibilities directly to the robot controller. The solution by Reis in contrast
relies on an additional safety controller monitoring robot operation. This results
in a redundant control architecture. The different concepts enable a worker to
enter the robot’s workspace while the task is still in progress by slowing robot
operation to a safe motion speed.
For close forms of cooperation, it is essential not only to detect motion in
3D but also to have an exact surveillance of the worker body. A stereoscopic
approach for 3D tracking of the worker body was developed by Fraunhofer IPK
in a cooperation with Fraunhofer IPA within the project team@work (Krüger J
et al. 2005). The system is able to detect segments of skin within the images
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 75
(Fig. 3.17b) in order to find characteristic points of the body. Based on this skin
detection the photogrammetric 3D capturing of the body is processed.
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
(a) Live image.
(b)surveillance
Detectedofrelevant
areas.
Fig. 23. Stereo camera based
human–robot-cooperation,
live image
(a) and detected relevant areas (b) [48].
635
Fig. 25. Impa
Figura 3.17: Stereo camera based surveillance of human-robot-cooperation.
are shown, bu
measured as w
tion during an
measure to qu
accidents and
Schraft extend
extremities ap
gles as input, and to generate a signal
output systems
which– robot
can integrated
switch between
no
Injury Index (H
3.4.2. as
Post-collision
sensors and light
Pubic Symphy
weight structures
risk, warn and stop (Winkler B, 2007).
showing thei
Joint torque sensing, together with a good robot model are used
The acquired distance data fromwithin
the PMD
is fast
being
preprocessed,
the by
robotics.
the LWRcamera
software for
detection
of collision or failure
Fig. 26 sho
an integrated torque observer. Inputs to the observer are the joint
foreground
extracted,
rectified and meshed
in Cartesian
to get
theandsurface
Fig. 25. Impactspace
tests performed
by DLR
Fraunhofer IPA [42,52–54].
Fig. 23. Stereo
camera based surveillance
of human–robot-cooperation,
live image
types of robot
torques
and motor positions. In order to indicate the resulting level
(a) and detected relevant areas (b) [48].
of the objects residing inside the workspace.
Additionally,
thewere
robot
is being
typical severit
of injury, so-called
severity indices
evaluated.
In the
impact of a hu
following section the results of the Head Injury Criterion (HIC)
modelled online in Cartesian space. By comparing both object representations,
threat
to the h
are shown, but other indices for the head, neck, and chest
were
Fig. 24 shows how a person whose arm is intruding into the
robot
seen from
the
camera,
the robot
can
measured
as well.
Thebe
HICidentified
evaluates theinside
resulting head acceleradynamic
safety model
area of theand
robotobjects
can be detected
with this
system.
during an impact. It is the most prominent and widely used
Krüger et al. [50] developed PMD camera based methods for
the camera object data. All no-robot-data is tion
treated
as obstacle and evaluated
measure to quantify the injury level of human beings caused by car
human motion analysis, which can be applied for workspace
against
safety
surrounding
the
accidents and was introduced to robotics in [51,52].Oberer and
surveillance
as well
as forareas
synchronisation
of human
androbot.
robot
Schraft extended these tests to impacts of head, chest and lower
motion in cooperative assembly tasks.
Fig. 3.18 shows how a person whose arm isextremities
intruding
into the dynamic safety
applying industrial robots (Fig. 25). The resulting Head
Injury Index (HIC), the Viscous Criteria (VC) for the chest and the
3.4.2. Post-collision
systems
– robot
sensorswith
and light
area of the
robot
canintegrated
be detected
this system.
Pubic Symphysis Peak Force (PSPF) for the pelvis are discussed,
weight structures
Krüger
al. (2007)
developed
camera
based
for limitations
human moshowing
theirmethods
potential and
for the situation in
Joint torque
sensing,et
together
with a good
robot modelPMD
are used
robotics.
within the
LWR
software
for
fast
detection
of
collision
or
failure
by
tion analysis, which can be applied for workspace surveillance as well as for synFig. 26 shows the results of HIC-measurements for different
an integrated torque observer. Inputs to the observer are the joint
of In
human
and robot
motion
in cooperative
types of robotsassembly
and impact tasks.
velocities up to 2 m/s. By means of
torqueschronisation
and motor positions.
order to indicate
the resulting
level
typical severity indices from automobile crash testing even an
of injury, so-called severity indices were evaluated. In the
impact of a huge robot such as the KR500 cannot pose a significant
following section the results of the Head Injury Criterion (HIC)
Post-collision systems
threat to the human head.
Fig. 24 shows how a person whose arm is intruding into the
dynamic safety area of the robot can be detected with this system.
Krüger
et al.of
[50]
developed PMDiscamera
basedcamera
methods for
Another system that enables close
forms
cooperation
a PMD
human motion analysis, which can be applied for workspace
based flexible workspace surveillance
systemasdeveloped
by Fraunhofer
IPA.and
The
surveillance
well as for synchronisation
of human
robot
in cooperative
tasks.
concept of this system is to use the motion
cameras
distanceassembly
data and
the robot joint an-
Robot integrated sensors and light weight structures
Joint torque sensing, together with a good robot model are used within the LWR
software for fast detection of collision or failure by an integrated torque observer.
Inputs to the observer are the joint torques
and motor positions. In order to indiFig. 24. Detection of a persons intruding arm: triangles marked in red are located
inside the yellow safety area [49].
cate the resulting level of injury, so-called severity indices were evaluated. In the
following section the results of the Head Injury Criterion (HIC) are shown, but
Fig. 26. Resulting
according to the E
76
extremities applying industrial robots (Fig. 25)
Injury Index (HIC), the Viscous Criteria (VC) fo
3.4.2. Post-collision systems – robot integrated sensors and light
Pubic Symphysis Peak Force (PSPF) for the p
weight structures
showing their potential and limitations fo
Joint torque sensing, together with a good robot model are used
robotics.
within the LWR software for fast detection of collision or failure by
Fig. 26 shows the results of HIC-measure
an integrated torque observer. Inputs to the observer are the joint
types of robots and impact velocities up to 2
torques and motor positions. In order to indicate the resulting level
typical
severity indices from automobile cra
of injury, so-called severity CAPITOLO
indices were evaluated.
the
3. THEInARTICLE
SELECTED
impact of a huge robot such as the KR500 canno
following section the results of the Head Injury Criterion (HIC)
threat to the human head.
26. Resulting
HIC36 values at varying impact veloci
Fig. 24.
of a persons
intruding arm:arm:
trianglestriangles
marked in red are
located
Figura 3.18: Detection
ofDetection
a persons
intruding
marked
inFig.
red
are located
according to the EuroNCAP Assessment Protocol and Biom
inside the yellow safety area [49].
inside the yellow safety area.
J. Krüger
et al.
/ CIRP Annals
- Manufacturing
Technology
58 (2009)
628–646
635
other indices for
the
head,
neck,
and chest
were
measured
as well. The HIC evaluates the resulting head acceleration during an impact. It is the most prominent
and widely used measure to quantify the injury level of human beings caused by
car accidents and was introduced to robotics in (Bicchi A et al. 2004, Haddadin
S et al. 2004). Oberer and Schraft extended these tests to impacts of head, chest
and lower extremities applying industrial robots (Fig. 3.19). The resulting Head
Injury Index (HIC), the Viscous Criteria (VC) for the chest and the Pubic Symphysis Peak Force (PSPF) for the pelvis are discussed, showing their potential and
limitations
for 628–646
the situation in robotics.
r et al. / CIRP Annals - Manufacturing
Technology 58 (2009)
635
(a)
Fig. 23. Stereo camera based surveillance of human–robot-cooperation, live image
(a) and detected relevant areas (b) [48].
(b)
Fig. 25. Impact tests performed by DLR and Fraunhofer IPA [42,52–54].
Figura 3.19: Impact tests performed by DLR and Fraunhofer IPA.
are shown, but other indices for the head, neck, and chest were
measured as well. The HIC evaluates the resulting head acceleration during an impact. It is the most prominent and widely used
measure to quantify the injury level of human beings caused by car
accidents and was introduced to robotics in [51,52].Oberer and
Schraft extended these tests to impacts of head, chest and lower
extremities applying industrial robots (Fig. 25). The resulting Head
Injury Index (HIC), the Viscous Criteria (VC) for the chest and the
3.4.2. Post-collision systems – robot integrated sensors and light
Pubic Symphysis Peak Force (PSPF) for the pelvis are discussed,
weight structures
showing their potential and limitations for the situation in
Joint torque sensing, together with a good robot model are used
Fig.for
25.fast
Impact
tests performed
by DLR
Fraunhofer
[42,52–54].
ot-cooperation,
withinlive
theimage
LWR software
detection
of collision
orand
failure
by IPArobotics.
Fig. 26 shows the results of HIC-measurements for different
an integrated torque observer. Inputs to the observer are the joint
types of robots and impact velocities up to 2 m/s. By means of
torques and motor positions. In order to indicate the resulting level
typical severity indices from automobile crash testing even an
of injury, so-called severity indices were evaluated. In the
are shown, but other indices for the head, neck, and chest were
is intruding into the
impact of a huge robot such as the KR500 cannot pose a significant
following section the results of the Head Injury Criterion (HIC)
measured as well. The HIC evaluates the resulting head acceleracted with this system.
threat to the human head.
tion during an impact. It is the most prominent and widely used
a based methods for
Fig. 24 shows how a person whose arm is intruding into the
dynamic safety area of the robot can be detected with this system.
Krüger et al. [50] developed PMD camera based methods for
human motion analysis, which can be applied for workspace
surveillance as well as for synchronisation of human and robot
motion in cooperative assembly tasks.
Fig. 3.20 shows the results of HIC-measurements for different types of robots
and impact velocities up to 2 m/s. By means of typical severity indices from
automobile crash testing even an impact of a huge robot such as the KR500 cannot
pose a significant threat to the human head.
n systems – robot integrated sensors and light
nsing, together with a good robot model are used
oftware for fast detection of collision or failure by
que observer. Inputs to the observer are the joint
r positions. In order to indicate the resulting level
led severity indices were evaluated. In the
the results of the Head Injury Criterion (HIC)
extremities applying industrial robots (Fig. 25). The resulting Head
Injury Index (HIC), the Viscous Criteria (VC) for the chest and the
Pubic Symphysis Peak Force (PSPF) for the pelvis are discussed,
showing their potential and limitations for the situation in
robotics.
Fig. 26 shows the results of HIC-measurements for different
types of robots and impact velocities up to 2 m/s. By means of
typical severity indices from automobile crash testing even an
impact of a huge robot such as the KR500 cannot pose a significant
threat to the human head.
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 77
persons intruding arm: triangles marked in red are located
ty area [49].
Fig. 26. Resulting HIC36 values at varying impact velocities for all robots, rated
Figura 3.20: Resulting
HIC36
at varying
velocities
for all robots, rated acaccording
to the values
EuroNCAP Assessment
Protocolimpact
and Biomechanical
Limits [54].
cording to the EuroNCAP Assessment Protocol and Biomechanical Limits
(Haddadin S et al. 2007).
3.1.4
Component supply and logistics
3.1.4.1
The materials supply challenge
As shown in Section 3.1.3 the developments in robotic assist and handling technology as well as safety systems increase efficiency of cooperative and automated assembly. However the overall time and cost efficiency of assembly processes does not only depend on the optimal relation between human and automation system but also on the question, how assembly parts are fed to cooperative workplaces. As human-machine cooperation targets to higher flexibility and
adaptability, the parts feeding process also needs to be highly flexible.
Several unpublished studies performed in Norwegian medium size manufacturing industries have revealed that much of the manual work effort in assembly
is not in the primary assembly tasks. The task of fetching components from storage areas and bringing them to the workplace for assembly has in many cases
been found to constitute 60-70% of the total work hours spent on assembly.
In automatic assembly solutions it has been found that more than 75% of the
equipment cost is on feeders and transport systems that automates the logistics
of assembly while the actual assembly operation equipment account for less than
20% of the total cost of a typical assembly line.
Nevertheless, much of the research in assembly automation has been focused
on the primary assembly. The automation of logistics and feeding has not been
studied to the same level. But this area needs much higher attention if the performance of automatic assembly shall be brought to the same high level as that
found in parts manufacturing. Some interesting papers by Bley, Denkena, Jovane
et al. has pointed out the challenges and opportunities for greater effectiveness
and flexibility in automatic assembly (2003-2006).
78
CAPITOLO 3. THE ARTICLE SELECTED
The real challenge here is part feeding flexibility. Solutions to automation in
feeding and logistics have existed for a long time, dating back to Henry Ford’s
assembly line for his model T automobile. But the majority of these solutions are
rigid, mass production types. Where flexibility is called for, in particular where
the need is flexibility for manufacture to order with lot sizes down to one unit,
good solutions are missing.
3.1.4.2
Flexible gripping devices
The flexibility challenge starts already at the gripping stage. In a cooperative
assembly between human and machine, the worker normally carries out complex
handling operations, where the high senso-motoric abilities of the human hand
are needed. Robots are inherently very flexible, but their performance is often
limited by the ability to grip the object that shall be handled. Grippers that can
handle the variety from hard to soft and limp materials are not commonly in use
industrially, and the technology in this area is still developing.
In particular the problem of changing quickly between different gripping tasks
is a challenge. This problem encompasses both the robot movement program
and the gripper design. Much advanced work has been performed on universal
multi-finger grippers, but these solutions have not seen any applications in ordinary industrial tasks. Furthermore there is a challenge in size variations, how to
enable the same gripping device to switch between handling objects of millimeter
size and meter size.
•
•
•
•
Humanoid multi-finger grippers.
Special purpose gripping systems.
Gripping of limp, soft and “non-definite shape” objects.
Large size range gripping.
3.1.4.3
Intelligent gripping
Fig. 3.21 shows the spectrum of today’s gripping systems.
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
y and logistics
ply challenge
on 3 the developments in robotic assist and
s well as safety systems increase efficiency of
mated assembly. However the overall time
assembly processes does not only depend on
between human and automation system but
how assembly parts are fed to cooperative
an–machine cooperation targets to higher
bility, the parts feeding process also needs to
Fig. 27. Spectrum of today’s gripping systems [61].
Figura 3.21: Spectrum of today’s gripping systems (Wegener K, 2007).
ed studies performed in Norwegian medium
ndustries have revealed that much of the
assembly is not in the primary assembly tasks.
omponents from storage areas and bringing
for assembly has in many cases been found to
the total work hours spent on assembly.
mbly solutions it has been found that more
ment cost is on feeders and transport systems
gistics of assembly while the actual assembly
feeding process also needs to
Fig. 27. Spectrum of today’s gripping systems [61].
ormed in Norwegian medium
revealed that much of the
in the primary assembly tasks.
m storage areas and bringing
3.1.
COOPERATION
OF HUMAN AND MACHINES IN ASSEMBLY LINES 79
as in many cases been
found
to
hours spent on assembly.
it has been found that more
This figure is based on a classification of grippers with two criteria: flexibifeeders and transport systems
blyJ. while
the
assembly
lity
and
complexity
of 58
a (2009)
gripper
and costs of a gripper. This spectrum has two
Krüger et
al. actual
/ CIRP Annals
- Manufacturing
Technology
628–646
s than 20% of the total cost of a
extreme positions: A two finger standard gripper is in the lower left corner of
s
Fig. 3.21.
h in assembly automation
has On the one hand a two finger standard gripper has only a very low
ly. The automation of
logistics but is on the otherFig.
flexibility,
a very
cost effective
gripping system.
28. Anthropomorph
hand [62].
the same level. But thisComplex
area
hands like the DLR hand II or the Schunk SAH (Schunk Anthroe performance
of automatic
opments
in robotic
assist
and
pomorphe
Hand) (Fig. 3.22) are very complex gripping systems with several
me
high level
as that
found in
systems
increase
efficiency
of
gripper. This spectrum has two extreme positions: A two finger
ing
papers bythe
Bley,
Denkena,
degrees
ly. However
overall
timeof freedom. But the price of these systems can easily exceed the price of
standard gripper is in the lower left corner of Fig. 27. On the one
allenges
and
opportunities
for
a robotonsystem.
esses does not only depend
hand a two finger standard gripper has only a very low flexibility,
nnautomatic
assembly
[55–60].
and automation system but
but is on the other a very cost effective gripping system.
eding
flexibility.
Solutions
to
y parts are fed to cooperative
• Sensory systems
for grippers.
Complex hands like the DLR hand II or the Schunk SAH (Schunk
have existed for a long time,
ooperation targets to higher
• Picking
unordered objects.
Anthropomorphe
Hand) (Fig. 28) are very complex gripping
mbly line for his model
T
feeding process also needs to
systems
with
several
degrees
of freedom.
But the[61].
price of these
ese solutions are rigid,•mass
Fig. in
27. gripping.
Spectrum
of today’s
gripping systems
Search mechanisms
systems can easily exceed the price of a robot system.
called for, in particular where
ormed in Norwegian medium
e to order with lot sizes down
e revealed that much of the
Sensory systems for grippers.
ing.
in the primary assembly tasks.
Picking unordered objects.
m storage areas and bringing
Search mechanisms in gripping.
as in many cases been found to
hours spent on assembly.
eady at the gripping stage. In a
In order to fill the gap between these two extreme gripping
it has been found that more
man and machine, the worker
systems
Fraunhofer IPA developed the IPA hand I and the IPA hand
feeders and transport systems
ng operations, where the high
II
(Fig.
29)
which is a good compromise between the flexibility and
mbly while the actual assembly
n hand are needed. Robots are
costs.
s than 20% of the total cost of a
rformance is often limited by
Both hands provide three possible gripping configurations (2
hall be handled. Grippers that
finger
parallel, 3 fingers lateral and 3 fingers centric), which enable
h in assembly automation has
oft and limp materials are not
to
grip
a large number of different objects. Furthermore, both
ly. The automation of logistics
Fig. 28. Anthropomorph
[62].
Figura
3.22:
Anthropomorph
handhand
(www.de.schunk.com).
the technology in this area is
hands
have
a limited
number of electric
drives
(the IPA hand I for
the same level. But this area
example just needs 2 drives), and thus they provide a good cost
he performance of automatic
ging quickly between different
effectiveness.
me high level as that found in
In
order
to
fill the
gap
between
these two
extreme
systems Fraunoblem encompasses both
the
The
IPA
hand
II is additionally
equipped
with gripping
aA
bio-inspired
gripper.
This
spectrum
has two extreme
positions:
two finger
ing papers by Bley, Denkena,
ripper design. Much hofer
advanced
finger
kinematics.
This
so-called
Finray
effect
enables
a
selfthe IPA
I and
IPAofhand
II On
(Fig.
standard gripper
is inhand
the lower
leftthe
corner
Fig. 27.
the 3.23)
one which is a
allenges and opportunities IPA
for developed
sal multi-finger grippers, but
adaptation
offinger
the fingers
to the
objects
that
have
to below
gripped
[61].
hand
a
two
standard
gripper
has
only
a
very
flexibility,
n automatic assembly
[55–60].
good compromise between the flexibility and costs.
ny applications in ordinary
but is on the other a very cost effective gripping system.
eeding flexibility. Solutions to
s a challenge in size variations,
Complex hands like the DLR hand II or the Schunk SAH (Schunk
have existed for a long time,
g device to switch between
Anthropomorphe Hand) (Fig. 28) are very complex gripping
embly line for his model T
and meter size.
systems with several degrees of freedom. But the price of these
ese solutions are rigid, mass
systems can easily exceed the price of a robot system.
s called for, in particular where
e to order with lot sizes down
Sensory systems for grippers.
ing.
efinite shape’’ objects.
Picking unordered objects.
Search mechanisms in gripping.
eady at the gripping stage. In a
In order to fill the gap between these two extreme gripping
man and machine, the worker
systems Fraunhofer IPA developed the IPA hand I and the IPA hand
day’s gripping systems.
ng operations, where the high
II (Fig. 29) which is a good compromise between the flexibility and
fication of grippers with two
n hand are needed. Robots are
costs.
Fig. 29.
IPA IPA
handshands
I and II I[61].
of a gripper and costs of a
Figura
3.23:
and II.
rformance is often limited by
Both hands provide three possible gripping configurations (2
hall be handled. Grippers that
finger parallel, 3 fingers lateral and 3 fingers centric), which enable
oft and limp materials are not
to grip a large number of different objects. Furthermore, both
Bothis hands
provide three possible gripping configurations (2 finger parallel,
the technology in this area
hands have a limited number of electric drives (the IPA hand I for
3 fingers lateral
and 3just
fingers
enable
to gripaagood
largecost
number of difexample
needs centric),
2 drives), which
and thus
they provide
ging quickly between different
effectiveness.
roblem encompasses both the
The IPA hand II is additionally equipped with a bio-inspired
ripper design. Much advanced
finger kinematics. This so-called Finray effect enables a selfrsal multi-finger grippers, but
adaptation of the fingers to the objects that have to be gripped [61].
ny applications in ordinary
s a challenge in size variations,
80
CAPITOLO 3. THE ARTICLE SELECTED
ferent objects. Furthermore, both hands have a limited number of electric drives
(the IPA hand I for example just needs 2 drives), and thus they provide a good
cost effectiveness.
The IPA hand II is additionally equipped with a bio-inspired finger kinematics. This so-called Finray effect enables a self adaptation of the fingers to the
objects that have to be gripped (Wengwner K, 2007).
3.1.4.4
General feeding subtasks
All automation solutions in manufacturing, whether it is in machining or cooperative assembly relies on some sort of automatic part feeding system to operate.
Feeding parts into an operation is essential for all manufacturing operations;
therefore automated as well as cooperative assembly requires automation of this
process. Automatic feeding has however proved to be rather challenging because
of the large diversity of parts. Even today there is no general theory or methodology that gives a straight road forward to an automation solution. There are on the
other hand large libraries of solutions that can be applied to specific subclasses
of the general class of automatic feeding.
To obtain an understanding of the feeding problem it is necessary to subdivide
the general task into typical task classes that depend on component characteristics as well as process characteristics. Langmoen (1983) and Rampersad (1994)
have proposed classification according to the following criteria:
• Production batch volume of a given unique component:
- Large (mass production)
- Medium
- Small and single piece (produce to order)
• Component size:
-
Large (characteristic dimension >1m)
Medium (characteristic size between 100 mm and 1 m)
Small (characteristic size between 1 mm and 100 mm)
Miniature (characteristic size < 1 mm)
• Component complexity:
- Simple shape, symmetric or semi symmetric
- Unsymmetrical but simple shape
- Complex shape
• Component stiffness:
- Stiff, non-deformable objects
- Semi stiff, deformable (like meat, rubber components, wires, etc.)
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 81
- Limp objects (like textile, tissues, thin plastic sheets, etc.)
• Fragility:
- Unbreakable by normal handling
- Moderately fragile (requires some attention to handling forces)
- Very fragile (must be handled very delicately).
This criteria list is not exhaustive, but it expresses the fact that very different
considerations have to be taken into account when a cooperative or automated
assembly system is considered. In the development of feeding systems such criteria can be used systematically to evaluate possible solutions and the likelihood
of obtaining an efficient, workable automatic feeding solution.
3.1.4.5
Design for assembly applied to feeding problems
Formalized methods of analysis with respect to feeding systems exist. Boothroyd
(2005) developed an analysis method as early as 1979. This method focuses mainly on geometrical properties of the part and assumes high volume production.
Later an alternative analysis method was proposed by Lien et al. (1982). In this
analysis more parameters are used so that stiffness, fragility and size are brought
into consideration. These two methods have both their shortcomings. This was
realized by Rampersad (1994) who has tried to combine the best elements from
Boothroyd and Lien into the “House of Assembly” method. The house of assembly also introduces some QFD elements in the analysis. Furthermore there is a
Japanese method for assembly automation analysis, the AREM (Suzuki T et al.
2001). This method relies on an expert evaluation of the automation solution.
Common to all of these methods is the idea that the complexity of feeding and
assembly can be characterized by some figure that expresses complexity of the
feeding task. In the Boothroyd analysis method which has been trademarked
DFMA there is in addition a calculation formula allowing the estimate of the cost
of feeding and assembling a component (Boothroyd G, 2005). This formula is
limited to certain cases of automatic assembly in the large volume segment.
3.1.4.6
Sub-problems of the feeding task
The feeding task can be subdivided into four major steps; each of them requiring
unique and very different techniques for automation. This subdivision assumes
that the components initially are stored in large numbers in bulk, not oriented
and fairly close, in the order of one meter or less, from the point of usage. Given
this border condition, the feeding task comprises the following steps according
to several authors (Rampersad HK 1994, Boothroyd G 2005, Estesen L et al. 19801982, Pherson D et al. 1984):
82
CAPITOLO 3. THE ARTICLE SELECTED
• Separation: One unique component is separated from the bulk volume.
• Transfer: The component is brought to a point very close to the point of pick
up or treatment in the next stage of the manufacturing operation.
• Orientation: The component is brought from a general orientation into the
specifically wanted orientation for the operation next in the process.
• Positioning: The component is positioned precisely within required tolerances for the next handling step in the process.
Some feeder techniques combine all these steps into one feeding device or
system. Such systems are the most common ones in large volume manufacturing
where small rigid parts are handled. For large parts, small volume, limp or fragile
parts it is more common to see these steps of feeding separated.
3.1.4.7
Feeders for small parts in large volume production
Boothroyd (2005) has described 17 basic feeder principles and mechanisms that
are suitable for large volume manufacturing. It is characteristic that all of these
combine mechanisms for the four subtasks of feeding into one device. It is common though that separation and transportation is related one part of the mechanism while orientation and positioning is relying on additional elements of the
system. In addition, the orientation techniques are in some respects generic
so that they may be applied in different feeding systems. Among the feeders
described by Boothroyd only a few have been widely put to use. These are:
•
•
•
•
vibratory bowl feeder,
elevator feeder,
belt feeder and
drum feeder.
Some of these feeder techniques are very old as a basic principle. The vibratory bowl feeder is one of them. Still it has the reputation for being the most used
feeder principle because of its simplicity and reliability. Over the last year some
more flexible and sophisticated feeders have emerged.
The basic belt feeder as described by Pherson et al. (1984) has been used as
basis for several developments that increase its flexibility and reconfigurability.
Pherson describes a system that uses simple photocell sensors to detect various
part orientations. Active elements like pneumatically operated wiper blades are
used to wipe away incorrectly oriented parts. The belt feeder is also the backbone
of systems that use camera based active orientation.
The need for gentle feeding for particularly sensitive parts has led to the development of the vibratory brush feeder (Fig. 3.24). This feeder has the advantage
of silent operation and small risk of surface damage of the part due to the part
falling into a bulk storage area.
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 83
638
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
limited range of shapes and dimensions. Som
in this category are:
sheet metal feeders for stacked sheet meta
tube feeders and bar feeders for raw ma
machining operations and
de-stackers for cardboard in box raising eq
These feeders are only usable for ma
surfaces like straight, un-machined bars a
material and cardboard and other large co
simple shapes [70–72]. As soon as the part b
something ‘‘un-straight’’ none of these fe
usable in their basic form. The feeding of m
thus has no general solutions available, nor do
any indication of theoretical solutions for the
challenges.
Fig. 30. Vibratory brush feeder for gentle feeding of fragile parts.
Figura 3.24: Vibratory
brush feeder for gentle feeding of fragile parts.
4.9. Orientation mechanisms
is an essen
system. Bringing the part into its required orie
Moving step elevators represent another variation of recently developed
feedfor therange
nextof
step
in and
almost
all manufac
limited
shapes
dimensions.
Some
operations.
Orientation
devices are often see
in ordinary
this category eleare:
ers. This elevator is also more gentle in its handling as compared to
of the feeder. But in almost all cases the orient
independent
from
the separation
an
vator feeders. But it has not had great impact in the feeder marketfunction
probably
due for
sheet metal
feeders
stacked
sheet meta
feeding.
oftube
feeders and bar feeders for raw mat
to its somewhat complicated feeding mechanism.
Basically there
are two
machining
operations
andprinciples used fo
638
Orientation
of the
component
J. Krüger et al. / CIRP Annals - Manufacturing Technology
58 (2009)
628–646
de-stackers for cardboard in box raising eq
In spite of many options for feeding, the vibratory bowl feeder
(Fig. orientation
3.25) is which utilizes the pote
Passive
energy
create
theonly
necessary
to
These to
feeders
are
usable force
for mat
the leading device by a solid margin. Its simplicity, versatility and reliability
are
incorrectly
surfaces
like oriented
straight, part.
un-machined bars a
the major reasons for its popularity. Despite this, it is one of the most
typical
“one usesandexternally
Active
orientation
suppli
material
and
cardboard
other large
com
reorient
or
eject
incorrectly
partsb
simple
As soonoriented
as the part
part only feeders” which means that it can only be used efficiently
forshapes
the [70–72].
spesomething ‘‘un-straight’’ none of these fee
Inorientation
addition
different
sensing
methods
cific part for which it is designed. This is mainly due to the fact that
usable
in theirtwo
basic
form. The
feeding
of m
part’s
orientation:
thus
has
no general solutions available, nor do
Fig. 31. Vibratory bowl feeder with active sensing reorientation. The vision system
devices are strongly
integrated
in
the
feeder
design.
identifies the component for pick up. Unsuitably positioned components are
any indication of theoretical solutions for the
returned to the bowl by a pneumatic pusher.
Fig. 30. Vibratory brush feeder for gentle feeding of fragile parts.
Figura 3.25:
Intrinsic sensing, that is utilization of geom
challenges.
part to create the reorientation or sorting ac
associated
with
passive orientation but can
4.9.
Orientation
mechanisms
In spite of many options for feeding, the vibratory bowl feeder
be combined with active orientation. It
(Fig. 31) is the leading device by a solid margin. Its simplicity,
available
for sensing
with passive
Orientation
of the component
isorientatio
an essen
versatility and reliability are the major reasons for its popularity.
external
energy
supply.
system.
Bringing
the
part into its required orie
Despite this, it is one of the most typical ‘‘one part only feeders’’
External
sensing
some
of sen
for
the next
step utilizing
in almost
allsort
manufac
which means that it can only be used efficiently for the specific part
observe features
thatdevices
indicates
orient
operations.
Orientation
arethe
often
seen
for which it is designed. This is mainly due to the fact that
supplyBut
information
obser
of can
the feeder.
in almost allfrom
cases this
the orienta
orientation devices are strongly integrated in the feeder design.
orientation
device.from
Mosttheactive
orientat
function
independent
separation
and
of external
feeding. sensing.
4.8. Large part feeding
Basically there are two principles used for
4.9.1. Passive orientation
The feeding of large parts seems to be a much more complex
Designing
passivewhich
orientation
to a
Passive
orientation
utilizesrelies
the pote
task than feeding of smaller parts. While small part feeders can
experience
the the
designer.
But force
Boothroy
energy to of
create
necessary
to r
accept parts in disordered bulk fashion large part feeding systems
library
of passive
elements [
incorrectly
orientedorientation
part.
cannot operate the same way.
the way uses
that externally
each element
s
atomic
Active in
orientation
supplie
The borderline between small part and large part feeding is not
orientation
(like reorientation
aro
reorient or effect
eject incorrectly
oriented parts.
very clear. But some indications about the borderline can be found
wanted reorientation device for a specific pa
in the rule of thumb saying that a vibratory bowl feeder should
by In
selecting
device
elemen
additionthe
twoorientation
different sensing
methods
have
a bowl diameter of approximately 10 times the largest
wanted
effect based on the available e
part’s
orientation:
Fig. 31. Vibratory bowl feeder with active sensing reorientation. The vision system
Vibratory
bowl component
feeder
active
sensing
reorientation.
The
visionof system
dimension
the partwith
to pick
feed.
For
practical
reasons
bowl feeders
features
the part. Edwardsen [73] has
identifies the of
for
up.
Unsuitably
positioned
components
are
returned
to
the
bowl
by
a
pneumatic
pusher.
are the
limited
in size to around
1 mup.
bowl
diameter, setting
the
bysensing,
investigating
the properties
of
further
Intrinsic
that is utilization
of geome
identifies
component
for pick
Unsuitably
positioned
components
are
borderline between small and large parts to 100 mm.
devices
and combining
that with
the mo
part to create
the reorientation
or sorting
ac
returned
to
the
bowl
by
a
pneumatic
pusher.
In other studies [68] the borderline has been set to 150–
energy
available
from a orientation
vibratory feeder
associated
with passive
but cant
spite
of there
manyisoptions
forindication
feeding, the
bowldefined
feeder
200Inmm.
But
no clear
thatvibratory
this is a well
precise
prediction
of active
the reorientation
be combined
with
orientation. effe
It
(Fig. 31) is Itthe
leading
device
a solid
margin.
borderline.
seems
more
to be by
a type
of border
setIts
bysimplicity,
the upper
Nevertheless,
all passive
orientation
metho
available
for sensing
with passive
orientatio
versatility
andavailable
reliability
are the
for its popularity.
size
limit for
feeders
formajor
bulk reasons
components.
geometrical
elements
that are not easily rec
external energy
supply.
Despite
is one
ofbe
the
most
typical ‘‘one having
part only
feeders’’
Trulythis,
largeitparts
can
seen
as components
at least
one
is utilizing
possible, some
the method
usu
reconfiguration
External sensing
sort of sens
which means
that it can
bemm.
used For
efficiently
for theno
specific
part
dimension
greater
thanonly
500
such parts
universal
effort
withfeatures
the associated
time for
reconfigu
observe
that indicates
the
orienta
for which
it is designed.
mainlyaredue
to the numerous
fact that
feeding
principles
seem toThis
exist.is There
however
passive
orientation
is still from
a method
main
can supply
information
this observ
orientation
devices
are strongly
integrated
in the designed
feeder design.
special
purpose
feeding
mechanisms,
usually
for a
production
automation.
orientation
device. Most active orientati
external sensing.
4.8. Large part feeding
4.9.1. Passive orientation
The feeding of large parts seems to be a much more complex
Designing passive orientation relies to a
84
CAPITOLO 3. THE ARTICLE SELECTED
3.1.4.8
Large part feeding
The feeding of large parts seems to be a much more complex task than feeding
of smaller parts. While small part feeders can accept parts in disordered bulk
fashion large part feeding systems cannot operate the same way.
The borderline between small part and large part feeding is not very clear. But
some indications about the borderline can be found in the rule of thumb saying
that a vibratory bowl feeder should have a bowl diameter of approximately 10
times the largest dimension of the part to feed. For practical reasons bowl feeders
are limited in size to around 1 m bowl diameter, setting the borderline between
small and large parts to 100 mm.
In other studies (Estensen L, 1980) the borderline has been set to 150-200 mm.
But there is no clear indication that this is a well defined borderline. It seems
more to be a type of border set by the upper size limit for available feeders for
bulk components.
Truly large parts can be seen as components having at least one dimension
greater than 500 mm. For such parts no universal feeding principles seem to
exist. There are however numerous special purpose feeding mechanisms, usually designed for a limited range of shapes and dimensions. Some examples of
feeders in this category are:
• sheet metal feeders for stacked sheet metal,
• tube feeders and bar feeders for raw material to sawing and machining
operations and
• de-stackers for cardboard in box raising equipment.
These feeders are only usable for materials with smooth surfaces like straight,
un-machined bars and tubes, flat plate material and cardboard and other large
component with clean simple shapes (Li HF et al. 2002, Liao X et al. 2003). As
soon as the part becomes formed into something “un-straight” none of these feeding principles are usable in their basic form. The feeding of more complex shapes
thus has no general solutions available, nor does the literature give any indication
of theoretical solutions for these large part feeding challenges.
3.1.4.9
Orientation mechanisms
Orientation of the component is an essential part of a feeder system. Bringing
the part into its required orientation is a necessity for the next step in almost
all manufacturing or assembly operations. Orientation devices are often seen as
integrated parts of the feeder. But in almost all cases the orientation can be seen
as a function independent from the separation and transportation part of feeding.
Basically there are two principles used for orientation (Boothroyd G, 2005):
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 85
• Passive orientation which utilizes the potential and movement energy to
create the necessary force to reorient or eject an incorrectly oriented part.
• Active orientation uses externally supplied energy to either reorient or eject
incorrectly oriented parts.
In addition two different sensing methods are used to sense the part’s orientation:
• Intrinsic sensing, that is utilization of geometrical features of the part to
create the reorientation or sorting action. This is normally associated with
passive orientation but can also to some extent be combined with active orientation. It is the only method available for sensing with passive orientation
since it requires no external energy supply.
• External sensing utilizing some sort of sensing devices that can observe
features that indicates the orientation of the part and can supply information from this observation to an active orientation device. Most active
orientation devices rely on external sensing.
Passive orientation
Designing passive orientation relies to a large degree on the experience of the
designer. But Boothroyd (2005) has presented a library of passive orientation elements. The library is atomic in the way that each element shall have only one
orientation effect (like reorientation around one axis). The wanted reorientation
device for a specific part is then composed by selecting the orientation device elements that can give the wanted effect based on the available external geometrical
features of the part. Edwardsen (1995) has taken this as step further by investigating the properties of some reorientation devices and combining that with the
movement pattern and energy available from a vibratory feeder to create a basis
for precise prediction of the reorientation effect.
Nevertheless, all passive orientation methods are based on fixed geometrical
elements that are not easily reconfigured. So even if reconfiguration is possible,
the method usually requires manual effort with the associated time for reconfiguration. Therefore the passive orientation is still a method mainly suitable for
mass production automation.
Active assisted gravity reorientation
In certain cases a hybrid solutions combining active and passive techniques are
applied. The typical case is the use of gravity for reorientation, but some sort
of active mechanism is applied to make gravity work in the way wanted. The
simplest procedure is to push the part into a position where gravity will ensure
86
CAPITOLO 3. THE ARTICLE SELECTED
proper orientation when the part tips over an edge (Estensen L, 1980). Here the
part must be fed in a well defined manner so that only the outcomes that give
correct orientation after the act of gravity reorientation can occur.
Another more sophisticated method uses vision system in combination with
a robot having only 4 DOF that can pick up randomly oriented parts on a table
or conveyor. These parts should typically have a length/width ratio greater than
one to make the action easier. The robot’s two gripper fingers are equippedwith
small low-friction pivoting pads as grip area. When a part is lifted off its resting
surface by a grip outside its center of gravity it will turn so that its center of
gravity comes straight below the centerline between the gripping pads. A 900
reorientation around the X or Y axis is thus possible for a robot system with only
rotation around the z-axis as a controlled orientation possibility.
These two examples show that hybrid reorientation systems that combine
gravity with some sort of active repositioning can work efficiently in certain applications.
Active sensing for reorientation
• Single point external sensing: Point or single feature sensing, like micro-switches,
photocells, inductive or fluidic sensors identify one single feature of the
part. These devices are simple and reliable and can usually be applied to the
actuation of one simple reorientation action. There are numerous examples
of how these sensors are used to obtain a specific reorientation action. But
the application of such sensing devices does seldom lead to greater flexibility of the feeder. So basically these sensors are applied in cases where
intrinsic sensing is unreliable or cannot be applied.
• Full feature sensing: The full feature sensing relies on some kind of mapping
of the complete external shape of the part. The most well known method
in this area is the electronic camera which gives a two-dimensional image
of the part, usually named vision systems. Pugh described the basis of this
method in 1983. Since then much of the development in the field has been
concentrated on refinement of the methods and development of more cost
effective solutions. The general strong reduction in cost for electronic devices has been a major driving force for the strong increase in vision system
applications for part orientation.
Some of the solution described in the literature covers sorting out all unwanted orientations and returning them to a new feeding loop. This is a technique
applied to vibratory bowl, vibrating brush and belt feeders (Boothroyd G 2005).
A more efficient method is use the image information to realign the part when
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 87
that is possible. This will in general give a substantial increase of the part feeding
rate (Fig. 3.26).
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
639
gravity reorientation
ybrid solutions combining active and passive
d. The typical case is the use of gravity for
e sort of active mechanism is applied to make
ay wanted. The simplest procedure is to push
n where gravity will ensure proper orientation
er an edge [68]. Here the part must be fed in a
so that only the outcomes that give correct
ct of gravity reorientation can occur.
phisticated method uses vision system in
robot having only 4 DOF that can pick up
parts on a table or conveyor. These parts
e a length/width ratio greater than one to
sier. The robot’s two gripper fingers are
ow-friction pivoting pads as grip area. When
esting surface by a grip outside its center of
o that its center of gravity comes straight
Fig. 32. Camera equipped belt feeder for feeding and picking arbitrarily oriented
Figura
Camera
equipped belt feeder for feeding and picking arbitrarily oriented
between the gripping pads.
A 9083.26:
reorientacomponents.
axis is thus possible for a robot system with
components.
d the z-axis as a controlled orientation
points 3D coordinates relative to the cameras can be computed.
The basic principle was described by Pugh [74]. Later there are
les show that hybrid reorientation systems
numerous authors who have described refinements of this method
Two-dimensional
has some
in identifying
the orientawith some sort of active repositioning
can
[75].imaging
Today the method
works shortcomings
particularly well in cases
where
rtain applications.
sharp corners and edges are visible so that edge detecting methods
tion of complex three-dimensional
shapes. Several methods have been investican be utilized to identify ‘‘same point’’ in two images.
or reorientation
gated to overcome this.
The vision technology is slowly put to use in industry. Some
sensing: Point or single feature sensing, like
interesting applications have been reported like sorting of castings
otocells, inductive or fluidic •
sensors
identify
Light
sectioninmapping.
this
method thin
of based
light are projected on
automotive In
part
manufacturing
[76,77]stripes
and robot
of the part. These devices are simple and
automation of flow assembly lines [78]. But these applications
the
angles. The light stripes form a pattern that are similar
ally be applied to the actuation of
one part
simpleat oblique
also reveal some practical challenges in keeping the camera
. There are numerous examples of
how
these
systems
in the
rather
harsh
industrial
to the sections
one operative
would get
bysometimes
cutting the
part
along
the plane of the light
obtain a specific reorientation action. But the
environments.
projection.
Information about the angle of the light-plane combined with
sensing devices does seldom lead
to greater
der. So basically these sensors are applied in
4.10.
Flexible component
feeders
the position of points
on the two-dimensional
image of the light section enc sensing is unreliable or cannot be applied.
The full feature sensing relies on some
kind
ables
a ofcomputation
ofthe
theartthree-dimensional
points
The state of
in feeding is basically centered
on on
the the parts surface
mplete external shape of the part. The most
concept having a specialized feeder for each part. In recent
that which
are seen
in the light section. By interpolation between points on difd in this area is the electronic camera
research projects there has been a concentration on finding more
onal image of the part, usually named
vision
feeding
principles that
handle variantsimage
of partsof the part can be
ferent light flexible
sections
a complete
thewill
dimensional
described the basis of this method in 1983.
without mechanical rebuilding. In recent years such systems have
generated
A,in 1983).
of the development in the field
has been (Pugh
appeared
industrial applications. These systems work well with
finement of the methods and development of
certain
applications
[65,79,80].
•
Multi
camera
three-dimensional
vision. This technique requires at least two
solutions. The general strong reduction in
The basic principle of multiple part feeding is to use the simple
evices has been a major driving cameras.
force for the Bybelt
identification
ofby
the
same[69]
point
theseparator
part’s surface in two or
feeder introduced
Pherson
as a on
simple
sion system applications for part orientation.
mechanism that presents the component on a surface where it can
more imagesbe with
known position in space the points 3D coordinates relaeasily detected. Then a suitable detection mechanism can be
on described in the literature covers
applied to identify
parts andThe
determine
location and was described by
tivesorting
to the cameras
can beindividual
computed.
basicitsprinciple
ntations and returning them to a new feeding
orientation.
Pugh.
Later there
are
numerous
authors
described
refinements
ue applied to vibratory bowl, vibrating
brush
The grip
action
shown in Fig.
33 is a who
typical have
example
of a
. A more efficient method is use
the
image
gripping
of
one
type
of
part
among
others.
of this method (Harris L et al. 1993). Today the method works particun the part when that is possible. This will in
larly
tial increase of the part feeding rate
(Fig.well
32). in cases where sharp corners and edges are visible so that edge
imaging has some shortcomings in identifydetecting methods can be utilized to identify “same point” in two images.
complex three-dimensional shapes. Several
nvestigated to overcome this.
The vision technology is slowly put to use in industry. Some interesting ap-
ng. In this method thin stripes of light are
have
rt at oblique angles. Theplications
light stripes form
a been reported like sorting of castings in automotive part manilar to the sections one would get by cutting
ufacturing (Baumann R 1982, Jermann JP 1983) and robot based automation of
ane of the light projection. Information about
t-plane combined with the
position
of points lines (Redford AH, 1986). But these applications also reveal some
flow
assembly
ional image of the light section enables a
e three-dimensional points on the parts
een in the light section. By interpolation
n different light sections a complete the
of the part can be generated [74].
dimensional vision. This technique requires at
y identification of the same point on the part’s
Fig. 33. Adept Flexfeeders for multi component feeding using electronic vision for
component identification.
ore images with known position in space the
e a length/width ratio greater than one to
asier. The robot’s two gripper fingers are
low-friction pivoting pads as grip area. When
resting surface by a grip outside its center of
so that its center of gravity comes straight
between the gripping pads. A 908 reorientaY axis is thus possible for a robot system with
88
nd the z-axis as a controlled
orientation
Fig. 32. Camera equipped belt feeder for feeding and picking arbitrarily oriented
components.
CAPITOLO
3. be
THE
ARTICLE SELECTED
points 3D coordinates relative
to the cameras can
computed.
The basic principle was described by Pugh [74]. Later there are
ples show that hybrid reorientation systems
numerous authors who have described refinements of this method
y with some sort of active repositioning can
the method works particularly well in cases where
practical challenges[75].
inToday
keeping
the camera systems operative in the sometimes
ertain applications.
sharp corners and edges are visible so that edge detecting methods
can beenvironments.
utilized to identify ‘‘same point’’ in two images.
rather harsh industrial
or reorientation
The vision technology is slowly put to use in industry. Some
l sensing: Point or single feature sensing, like
interesting applications have been reported like sorting of castings
3.1.4.10
Flexibleincomponent
feeders
otocells, inductive or fluidic
sensors identify
automotive part manufacturing [76,77] and robot based
of the part. These devices are simple and
automation of flow assembly lines [78]. But these applications
ually be applied to the actuation
of one simple
alsoinreveal
some is
practical
challenges
in keeping
The state
of the art
feeding
basically
centered
on the
thecamera
concept having a spen. There are numerous examples of how these
systems operative in the sometimes rather harsh industrial
cialized
each part. In recent research projects there has been a conobtain a specific reorientation
action.feeder
But the for
environments.
sensing devices does seldom lead to greater
centration on finding more flexible feeding principles that will handle variants of
eder. So basically these sensors are applied in
4.10. Flexible component feeders
sic sensing is unreliable or
cannot
be applied.mechanical rebuilding. In recent years such systems have appeared
parts
without
The full feature sensing relies on some kind of
The state of the art in feeding is basically centered on the
intheindustrial
applications.
These systems work well with certain applications
mplete external shape of
part. The most
concept having a specialized feeder for each part. In recent
d in this area is the electronic
camera
which
research
projects
there1986,
has been
a concentration
finding more
(Boothroyd G 2005, Redford AH
Suzuki
et al. on
1980).
sional image of the part, usually named vision
flexible feeding principles that will handle variants of parts
The basic
of multiple
partInfeeding
tosystems
use the
described the basis of this method
in 1983.principle
without mechanical
rebuilding.
recent yearsis
such
havesimple belt feeder
of the development in the field has been
appeared in industrial applications. These systems work well with
introduced by Pherson (1984) as a simple separator mechanism that presents the
finement of the methods and development of
certain applications [65,79,80].
e solutions. The generalcomponent
strong reductionon
in a surface
The basic
principle
multiple
part feeding
is to use the
simple
where
it of
can
be easily
detected.
Then
a suitable detection
devices has been a major driving force for the
belt feeder introduced by Pherson [69] as a simple separator
can bemechanism
appliedthat
topresents
identify
individual parts and determine its location
ision system applicationsmechanism
for part orientation.
the component on a surface where it can
and orientation. be easily detected. Then a suitable detection mechanism can be
ion described in the literature covers sorting
applied to identify individual parts and determine its location and
The
grip
action
shown in Fig. 3.27 is a typical example of
ntations and returning them to
a new
feeding
orientation.
que applied to vibratory bowl, vibrating brush
The grip action shown in Fig. 33 is a typical example of a
type of part among
others.
]. A more efficient method is use the image
gripping of one type of part among others.
n the part when that is possible. This will in
ntial increase of the part feeding rate (Fig. 32).
imaging has some shortcomings in identifyf complex three-dimensional shapes. Several
nvestigated to overcome this.
a gripping of one
ing. In this method thin stripes of light are
art at oblique angles. The light stripes form a
milar to the sections one would get by cutting
lane of the light projection. Information about
ht-plane combined with the position of points
sional image of the light section enables a
he three-dimensional points on the parts
een in the light section. By interpolation
n different light sections a complete the
of the part can be generated [74].
dimensional vision. This technique requires at
By identification of the same point on the part’s
33. Adept Flexfeeders for multi component feeding using electronic vision for
Figura 3.27: Adept Fig.
Flexfeeders
for multi component feeding using electronic
component identification.
more images with known position in space the
vision for
component identification.
Laboratory experiments in the NTNU assembly laboratory as shown in Fig.
3.28 have shown though that the time required to identify one single component
among several different ones can take considerable time.
The time for the automatic picking procedure will be longer than what can
be obtained with dedicated feeder and orientation systems. But the advantage is
that a changeover between parts can be performed very quickly, there is hardly
any need for hardware changes. And for previously used components the software templates will be available and immediately applicable. The main challenge
in flexible feeding seems to be the time required to empty a feeder of one type and
replenish it with another type of component.
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 89
640
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
Instead of this the transport vehicle use
systems that rely on observing fixed la
they use information transmitted from
move to get overall information ab
systems require much less of costly fixe
thus much easier to deploy and reconfi
4.11.2. Intelligent conveyor systems
Intelligent conveyor systems are ot
opments. Conveyors have gone from
with a constant conveying speed to be
tation systems that can route individu
are tailored to the needs of a specific m
routes available will always be limite
tracks laid down, but within these there
greater flexibility than in the old rigid
Systems frequently use some sort o
transport bases (pallets) to determ
individual transport unit. It is thus
products in different stages of the p
transport system. Each individual trans
its destination at each conveyor inte
production process for this part as r
Fig. 34. Camera and force controlled gripper fingers used to pick up unordered
control
system of the conveyor.
Figura 3.28: Camera
and force controlled gripper fingers used to pick up
unordered
castings [Fanuc].
This type of transport enables the c
castings [Fanuc].
automatic stations along a core transp
combination of serial and parallel wo
obtain the most efficient utilization of
Laboratory experiments in the NTNU assembly laboratory as
the automatic machines in hybrid assem
shown in Fig. 34 have shown though that the time required to
Another wayidentify
of looking
at component
flexible feeding
is todifferent
eliminate
feeding
[4]. The
use of manual workers for the
one single
among several
ones the
can special
assembly
take
considerable
time. The system
time forequipped
the automatic
picking
device altogether
and
rely on sensory
robots
for the
heavylines
andgive both greater capa
overall productivity than fixed sequ
procedure will be longer than what can be obtained with dedicated
monotonous task
of picking components either from conveyors or out
of storage
without the possibility for intelligent r
feeder and orientation systems. But the advantage is that a
changeover between
can besystems
performed show
very quickly,
there is
bins. Here developments
of binparts
picking
promises.
4.12. Intelligent warehousing
hardly any need for hardware changes. And for previously used
Fanuc has introduced
a system that combines an arm mounted camera system
components the software templates will be available and
One challenge
immediately
applicable.that
The enables
main challenge
in flexible
feeding
with force sensors
in the grippers
the robot
to pick
parts directly
from in assembly is to ke
stock at all storage points in a system. T
seems to be the time required to empty a feeder of one type and
a container or from
a conveyor.
large when the process is automated. O
replenish it with another type of component.
point lead to immediate stop of ope
Another way of looking at flexible feeding is to eliminate the
manual intervention to replenish and
special feeding device altogether and rely on sensory system
3.1.4.11 Flexible
automated logistics
Today we see very little of true auto
equipped robots for the heavy and monotonous task of picking
to automated assembly systems. Bu
components either from conveyors or out of storage bins. Here
Advanced automated
guided
systems
assembly flow control system that
developments
of binvehicle
picking systems
show promises.
components
Fanuc
has
introduced
a
system
that combines
an armAutomatic
mounted
The feeding of component feeders is another
challenge.
operationinofthe system and orders r
points is needed for a system to opera
camera system with force sensors in the grippers that enables the
assembly systems
requires
a steady
flowa container
of components
from intermediate
sto-systems are well devel
Today’s ERP
robot
to pick parts
directly from
or from a conveyor.
bookkeeping
rage or part manufacture directly before assembly. Traditionally this section of
of the main storage are
materials, intermediate stock and fini
4.11. Flexible automated logistics
the manufacturing system has been automated to a lesser degree since
level ofthe
any freof these storage areas is no
tight
down
to the minute follow up of p
4.11.1.
Advanced
automated
guided
vehicle
systems
quency of operations is lower, and consequently the apparent economic justificathus unable to provide exact informati
The feeding of component feeders is another challenge.
tion is harder toAutomatic
reach. operation of assembly systems requires a steady flow
in progress stock outside of the main st
kept track
off. Thus there is a large ‘‘whi
components
fromare
intermediate
storage
or part manufacture
But also in of
this
area there
technical
developments
that points
towards
in most factories. The problem of this ‘‘
directly before assembly. Traditionally this section of the
more flexible automatic
transport
solutions.
Automated
(AGV)
most foremen
and department manag
manufacturing
system has
been automated
to a lesserguided
degree vehicle
storekeeping
to cover the needs of any
frequency
of operations
lower, and consequently
the
technology has since
beenthe
used
for many
years,isparticularly
in warehousing
and large
The developments in new sensor te
apparent economic justification is harder to reach.
scale manufacturing
(Bostelman RV et al. 2005, Correa et al. 2007,development
Lin L et al.
But also in this area there are technical developments that
of RFID, offers great poten
more flexible
automatic transport
have real
2006, Maughanpoints
FG ettowards
al. 2000).
This technology
is nowsolutions.
ready forthat
a new
steptime information abou
Automated guided vehicle (AGV) technology has been used for
points [88–90]. These solutions should
forward with the
emergence of low cost advanced electronic systems
and aumany years, particularly in warehousing and large scale manuwhich is necessary for reliable automat
facturing [81–84]. This technology is now ready for a new step
and monitoring systems are not yet av
forward with the emergence of low cost advanced electronic
the demand from industry for a tighter,
systems and autonomous navigation systems that will make
in progress stock levels is increasin
modern AGVs more cost effective and flexible, and thus also
research program address this task in
applicable to a much wider range of assembly systems [85–87].
generation Manufacturing Execution Sy
In particular the emergence of trackless AGV systems bears
stock control. The emergence of such s
promises of more efficient and flexible autonomous transport
precise supply of components for assem
90
CAPITOLO 3. THE ARTICLE SELECTED
tonomous navigation systems that will make modern AGVs more cost effective
and flexible, and thus also applicable to a much wider range of assembly systems
(Yamada Y et al. 2003, Yoo J-W et al. 2005, Zhang Z et al. 2005).
In particular the emergence of trackless AGV systems bears promises of more
efficient and flexible autonomous transport devices. In trackless solutions no
fixed guide wires are needed. Instead of this the transport vehicle uses either onboard navigation systems that rely on observing fixed landmarks for navigation,
or they use information transmitted from cameras in the area they move to get
overall information about their position. These systems require much less of costly fixed guidance systems and are thus much easier to deploy and reconfigure.
Intelligent conveyor systems
Intelligent conveyor systems are other important new developments. Conveyors have gone from being simple belt systems with a constant conveying speed
to becoming complex transportation systems that can route individual objects
along paths that are tailored to the needs of a specific manufacturing process.
The routes available will always be limited to those offered by the tracks laid
down, but within these there are opportunities for much greater flexibility than
in the old rigid one-way systems.
Systems frequently use some sort of identification system on transport bases
(pallets) to determine the routing of each individual transport unit. It is thus
possible to mix different products in different stages of the production process
on one transport system. Each individual transport unit will be routed to its
destination at each conveyor intersection according to the production process for
this part as represented in the overall control system of the conveyor.
This type of transport enables the combination of manual and automatic stations along a core transport line. It also enables the combination of serial and parallel work stations in a process to obtain the most efficient utilization of the manual work force and the automatic machines in hybrid assembly lines as shown
by Lien (2001). The use of manual workers for the complex finishing tasks on assembly lines give both greater capacity flexibility and higher overall productivity than fixed sequential assembly systems without the possibility for intelligent
rerouting.
3.1.4.12
Intelligent warehousing
One challenge in assembly is to keep track of the amount in stock at all storage
points in a system. This challenge becomes very large when the process is automated. Out of stock situations at any point lead to immediate stop of operations,
and the need for manual intervention to replenish and restart.
Today we see very little of true automatic storekeeping applied to automated
assembly systems. But the need is there. An assembly flow control system that
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 91
keeps track of all the components in the system and orders replenishment to all
storage points is needed for a system to operate truly automatic.
Today’s ERP systems are well developed to handle the overall bookkeeping
of the main storage areas in a company, i.e. raw materials, intermediate stock and
finished goods. But the stock level of any of these storage areas is not updated in
real time. For tight down to the minute follow up of production the ERP system
is thus unable to provide exact information. In addition all the work in progress
stock outside of the main storage areas is normally not kept track off. Thus there
is a large “white spot” on the storage map in most factories. The problem of this
“white spot” is overcome by most foremen and department managers through
local notebook storekeeping to cover the needs of any department.
The developments in new sensor technology, in particular the development of
RFID, offers great potentials for building solutions that have real time information about the status of all storage points (Djassemi M et al. 2005, Ramamurthy
H et al. 2005, Ranky PG 2006). These solutions should have accuracy of one unit
which is necessary for reliable automatic operation. Such planning and monitoring systems are not yet available on the market. But the demand from industry
for a tighter, real time control of all work in progress stock levels is increasing. At
NTNU two ongoing research program address this task in an effort to develop
next generation Manufacturing Execution Systems (MES) with real time stock
control. The emergence of such systems will enable a more precise supply of
components for assembly and minimize time loss due to out of stock situations.
3.1.4.13
Parts feeding with intelligent assist devices
An efficient cooperation of human and machine in assembly can often be realized in a way where the feeding of the assembly parts is executed by an automated handling system but the assembly process itself is carried out by the human
worker. When a close interaction with contact of human and feeding system is
possible with respect to safety requirements, the robotic feeding system may not
only feed the assembly workpiece but also support the worker with positioning
functionality.
Fig. 3.29 shows a cobotic rail crane system design from Fraunhofer IPK, which
provides feeding of large assembly parts as well as support for positioning. For
the safe and efficient interaction, the intelligent automation device also provides
functionality for collision avoidance. Additionally, in order to give the worker free space for his work, the cobotic system after feeding returns to a defined
position executing the so-called homing function.
92
CAPITOLO 3. THE ARTICLE SELECTED
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
Fig. 35. Cobotic system for feeding and positioning of large assembly parts.
Figura 3.29: Cobotic
system for feeding and positioning of large assembly parts.
4.13. Parts feeding with intelligent assist devices
Fig. 36. Activities performed manually, automatically by
An efficient cooperation of human and machine in assembly can
by robots in the enterprises where robots are installed [
often be realized in a way where the feeding of the assembly parts
is executed by an automated handling system but the assembly
process itself is carried out by the human worker. When a close
interaction with contact of human and feeding system is possible
3.1.5.1 Requirements
fortoflexible
robot based
automation
with respect
safety requirements,
the robotic
feeding system
may not only feed the assembly workpiece but also support the
worker with positioning functionality.
Fig. 35 shows a cobotic rail crane system design from
The ARFLEX project
funded within the 6th Framework programme of the EU
Fraunhofer IPK, which provides feeding of large assembly parts
as well
as support
for positioning. of
Forathe
safe and industrial
efficient
aims to enhance and
extend
the capabilities
common
robot system.
interaction, the intelligent automation device also provides
By means of advance
control
system
and intelligent
sensor
ARFLEX disfunctionality
for collision
avoidance.
Additionally, in
order todevices
give
the
worker
free
space
for
his
work,
the
cobotic
system
after
feeding
places nonreconfigurable, proprietary and expensive automation technologies.
returns to a defined position executing the so-called homing
Kus et al. (2008)function.
analysed the requirements of small and medium enterprises
3.1.5
Organizational and economic aspects
Fig. 37. Main disadvantages of using the robot in indu
(SME). The investigation
was and
carried
outaspects
among enterprises in Germany,
Italy,
5. Organizational
economic
Poland and Switzerland. The analysis contained questions related to the robo5.1. Requirements for flexible robot based automation
5.2. Hybrid automation cost assessment
tisation of the production line, the activities performed manually, automatically
The ARFLEX project funded within the 6th Framework
40 shows the
by the use of specialized
machines, and by robots. Figs. 3.30a-3.30d Fig.
comprise
es-assumed cost potentials o
programme of the EU aims to enhance and extend the capabilities
compared with robotic workcells and autom
of a common
industrial robot
system.
By means
advance controlof robotic
sential results of the
investigation,
with
respect
to of
complexity
Accordingsystems,
to this, hybrid automation has th
system and intelligent sensor devices ARFLEX displaces nonbenefit in small and medium sized productio
production batch reconfigurable,
sizes and changes
lines of
SME.
proprietary in
andproduction
expensive automation
technoloA more specific way to asses the costs of h
gies. Kus et al. analysed the requirements of small and medium
the net present value (NPV), which is the sum o
specialized machines, and by robots. Figs. 36–39 comprise
consists of the following tasks:
The goal of the
enquiry
concerned
the identification
of the limitations
andback to its present value.
enterprises
(SME)
[91]. The investigation
was carried out among
outflow discounted
enterprises in Germany, Italy, Poland and Switzerland. The analysis
The following example shows how this me
imperfections of contained
the robotic
systems having the biggest influence
on their dequestions related to the robotisation of the production
to assess the costs of hybrid automation. This e
line,
the
activities
performed
manually,
automatically
by
the
use
of
PowerMate
example [10] shown in Section 2.3
ployment in the industrial applications.
essential results of the investigation, with respect to complexity
In case of small
enterprises the main hindrance is too small production batchof robotic systems, production batch sizes and changes in
production
lines
of SME.
es. This result indicates
that
the
market is still not served with solutions satisfying
The goal of the enquiry concerned the identification of the
the requirements limitations
of the low
volume production
In case
of the mediumand imperfections
of the robotic system.
systems having
the
biggest
influence on their
deployment
in the industrial
applicasize enterprises, the
deployment
of the
industrial
robots is
mainly limited by the
tions.
high costs.
In case of small enterprises the main hindrance is too small
production batches. This result indicates that the market is still not
served batch
with solutions
satisfying
the observed
requirements in
of electronics
the lowThe most frequent
changes
can be
and electrical
volume production system. In case of the medium-size enterprises,
equipment production.
In this
sector
almost
of limited
the batch
the deployment
of the
industrial
robots 60%
is mainly
by the changes consider
high
costs.
the daily changes in the production. Also in other industrial sectors the daily
The most frequent batch changes can be observed in electronics
changes in the production
are also
prevailing
andalmost
are more
and electricalline
equipment
production.
In this sector
60% of than 40%.
the batch changes consider the daily changes in the production.
Also in other industrial sectors the daily changes in the production
line are also prevailing and are more than 40%.
Fig. 38. The main obstacles for the dissemination of the
respect to the size of the enterprise [91].
Fig. 36. Activities performed manually, automatically by specialized machines and
e in assembly can
by robots in the enterprises where robots are installed [91].
he assembly parts
but the assembly
er. When a close
ystem is possible
Fig. 35. Cobotic system for feeding and positioning of large assembly parts.
c feeding system
al. / CIRP
Annals - the
Manufacturing Technology 58 (2009) 628–646
also
support
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 93
641
m design 4.13.
from Parts feeding with intelligent assist devices
e assembly parts
Fig. 36. Activities performed manually, automatically by specialized machines and
An efficient cooperation of human and machine in assembly can
by robots in the enterprises where robots are installed [91].
afe and efficient
often be realized in a way where the feeding of the assembly parts
e also provides
executed by an automated handling system but the assembly
y, in order toisgive
process itself is carried out by the human worker. When a close
tem after feeding
interaction with contact of human and feeding system is possible
o-called homing
with respect to safety requirements, the robotic feeding system
may not only
thedisadvantages
assembly workpiece
but also
support
the
Fig.feed
37. Main
of using the robot
in industrial
applications
[91].
worker with positioning functionality.
Fig. 35 shows a cobotic rail crane system design from
tion
Fraunhofer5.2.
IPK,
which
provides feeding
of large assembly parts
Hybrid
automation
cost assessment
as well as support for positioning. For the safe and efficient
6th Framework
interaction, the
intelligent
also
provides
Fig. 40
shows the automation
assumed costdevice
potentials
of hybrid
automation
d the capabilities
functionality
for
collision
avoidance.
Additionally,
in
order
to
give
compared
with
robotic
workcells
and
automated
transfer
lines.
arge assembly parts.
f advance control
the workerAccording
free spaceto
forthis,
his work,
theautomation
cobotic system
feedingeconomic
hybrid
has after
the biggest
X displaces returns
nontobenefit
a defined
position
executing
theproductions.
so-called homing
in small
and medium
sized
mation technolofunction.
A more specific way to asses the costs of hybrid automation is
mall and medium
Fig. 37. Main disadvantages of using the robot in industrial applications [91].
the net present value (NPV), which is the sum of all cash inflow and
Fig.(a)
36. Activities
performed
manually, automatically
by specialized
machines and (b) Main disadvantages of using the robot in
Activities
performed
manually,
automatically
5. Organizational
and economic
aspects
arried
out among
outflow
back
to its
present
value.
ne in assembly
can
by robots discounted
in the enterprises
where
robots
are installed
[91].
by specialized
robots
industrial applications.
and.assembly
The analysis
The following
examplemachines
shows howand
this by
method
canin
be the
applied
the
parts
ofbut
thethe
production
5.1. Requirements
for
based
automation
where
robots
are installed.
to assessenterprises
theflexible
costs ofrobot
hybrid
automation.
This example bases on
theHybrid automation cost assessment
5.2.
assembly
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2
642
cally
the use
of
PowerMate example [10] shown in Section 2.3. Thus, the assembly
rker. by
When
a close
36–39
comprise
The
ARFLEX
project
funded
within
the
6th
Framework
consists
of
the
following
tasks:
system is possible
Fig. 40 shows the assumed cost potentials of hybrid automation
ct to
complexity
programme of the EU aims to enhance and extend the capabilities
tic
feeding
system
compared with robotic workcells and automated transfer lines.
and
in
ut
alsochanges
support
of athe
common industrial robot system. By means of advance control
According to this, hybrid automation has the biggest economic
system and intelligent sensor devices ARFLEX displaces nonbenefit in small and medium sized productions.
tem design
from
ntification
of
the
reconfigurable,
proprietary and expensive automation technoloA more specific way to asses the costs of hybrid automation is
ge
assembly
parts
tems
having
theKus et al. analysed the requirements of small and medium
gies.
the net present value (NPV), which is the sum of all cash inflow and
safe
andapplicaefficient
dustrial
enterprises (SME) [91]. The investigation was carried out among
outflow discounted back to its present value.
ice also provides
enterprises in Germany, Italy, Poland and Switzerland. The analysis
The following example shows how this method can be applied
ly, inisorder
give
ance
too to
small
contained
questions related to the robotisation of the production
to assess the costs of hybrid automation. This example bases on the
ystem
after
feeding
market is still
notthe activities performed manually, automatically by the use of
line,
PowerMate example [10] shown in Section 2.3. Thus, the assembly
so-called
ents
of thehoming
lowspecialized
machines, and by robots. Figs. 36–39 comprise
consists of the following tasks:
-size enterprises,
essential results of the investigation, with respect to complexity
Fig. 37. Main disadvantages of using the robot in industrial applications [91].
ly limited by
of the
robotic systems, production batch sizes and changes in
production lines of SME.
ved in electronics
The goal of the enquiry concerned the identification of the
ation
5.2. Hybrid automation cost assessment
or almost 60%
of
limitations
and imperfections of the robotic systems having the
n the production.
biggest influence
onshows
their the
deployment
in the
industrial
applica6th Framework
Fig. 40
assumed cost
potentials
of hybrid
automation
in the production
tions.
Fig.
38. The main
obstacles
forworkcells
the dissemination
of the robotic
technology
with
nd the capabilities
Fig.
39.
frequency
of the changes
in the
productionin
linethe
(batch
change) with
compared
with
robotic
and
automated
transfer
lines.
(c)
The
main
obstacles
for
the
dissemination
of the
(d)TheThe
frequency
of the
changes
producto theenterprises
size of the enterprise
[91].hindrance is too small
In caserespect
of small
the automation
main
respect to the industrial sector [91].
of advance control
According
to this, hybrid
has the biggest economic
robotic
technology
with
to the
size
tion line (batch change) with respect to the
batches.
result
that respect
the
market
is still
notof the
EX displacesproduction
nonbenefit
inThis
small
and indicates
medium sized
productions.
enterprise.
industrial sector.
served with solutions
satisfying
the
requirements
the automation
lowomation technoloA more
specific
way to
asses
the costs of of
hybrid
is
Task 1: Prepare part A.
volume production
system.
In case
of the
medium-size
small and medium
the net present
value
(NPV),
which
is the sum enterprises,
of all cash inflow and
Task 2:of
Grip
part
B out of bin.
the deployment
of discounted
the industrial
is mainly
limited
by the
carried out among
3.30:
Essential
results
the
investigation.
outflow
backrobots
to itsFigura
present
value.
Task 3: Move B to A.
rland. The analysis
high costs.
The following example shows how this method can be applied
Task 4: Precise positioning B.
of the production
The most
batch
be observed
electronics
tofrequent
assess the
costschanges
of hybridcan
automation.
Thisinexample
bases onthe
Task 5: Assembly A and B to AB.
tically by theand
use electrical
of
PowerMate
example
[10] shown
in Section
2.3. Thus,
theofassembly
equipment
production.
In this
sector almost
60%
Task 6: Move part AB to part C.
36–39 comprise
consists of
the following
tasks:
the batch changes
consider
the daily
changes in the production.
pect to complexity
Also in other industrial sectors the daily changes in the production
Fig. 38. The main obstacles for the dissemination of the robotic technology with
and changes
in also prevailing and are more than 40%.
As to
mentioned
earlier
tasks
like handling of huge and heavy
line are
respect
the size of the
enterprise
[91].
objects can be more efficient with robots while assembly tasks
entification of the
where high sensory skills are required can be more efficient with
ystems having the
human beings. As consequence the tact times for a task differ if
industrial applicathey are done by a robot, by a human being or in a hybrid
automation scenario. Table 1 compares tact times of human and
rance is too small
robot for the task sequence example.
e market is still not
Furthermore, in a hybrid automation scenario human worker
ments of the lowand robot can work in parallel. In this scenario the task 1 can be
Fig. 40. Reduction o
m-size enterprises,
done by the human worker while the robot performs tasks 2 and 3.
assumed cost potent
nly limited by the
As consequence the tact time of the hybrid system reduces to 100 s
(instead of 140 s).
erved in electronics
The tact time diagrams (Figs. 41–43) for these three alternatives
ctor almost 60% of
look as follows.
in the production.
Assuming that the needed output is 75,000 parts AB per year
s in the production
Fig. 38. The main obstacles for the dissemination of the robotic technology with
and that this company is producing in 2 shifts à 8 h at 200 days per
.
respect to the size of the enterprise [91].
year, the company will need either (Table 2):
3.1.5.2
Hybrid automation cost assessment
Fig. 3.31 shows the assumed cost potentials of hybrid automation compared
with robotic workcells and automated transfer lines. According to this, hybrid
automation has the biggest economic benefit in small and medium sized productions.
A more specific way to asses the costs of hybrid automation is the net present
value (NPV), which is the sum of all cash inflow and outflow discounted back to
its present value.
The following example shows how this method can be applied to assess the
costs of hybrid automation. This example bases on the PowerMate example
(Schrft RD et al. 2005) shown in Section 3.1.2.3. Thus, the assembly consists
10 (9.6) workers (human solution) or
3 (2.2) robot systems (automated solution) or
1 hybrid system with 3 human workers and 1 robot.
Furthermore we assume the following cash inflows and
outflows for the investments.
According to these cash flows as the estimated numbers of
Fig
94
CAPITOLO 3. THE ARTICLE SELECTED
Figura 3.31: Reduction of robot system costs compared to labour cost (left) and assumed
cost potentials of future hybrid automation (right).
of the following tasks:
Task 1:
Task 2:
Task 3:
Task 4:
Task 5:
Task 6:
Prepare part A.
Grip part B out of bin.
Move B to A.
Precise positioning B.
Assembly A and B to AB.
Move part AB to part C.
As mentioned earlier tasks like handling of huge and heavy objects can be
more efficient with robots while assembly tasks where high sensory skills are
required can be more efficient with human beings. As consequence the tact times
for a task differ if they are done by a robot, by a human being or in a hybrid
automation scenario. Table 3.1 compares tact times of human and robot for the
task sequence example.
Tabella 3.1: Tact time comparison between human and robot.
Tact times in [s]
Robot: all tasks
sequentially
Human: all tasks
sequentially
1 and 2,3 parallel,
rest sequentially
Best of
Task 1:
Task 2:
Task 3:
Task 4:
Task 5:
Task 6:
40
20
20
60
60
20
40
80
120
30
30
20
40
20
20
20
20
20
40
20
20
30
30
20
220
320
100
160
Prepare A
Grip B
Move B to A
Precise positioning B
Assembly
Move AB to C
Total tact time
Furthermore, in a hybrid automation scenario human worker and robot can
work in parallel. In this scenario the task 1 can be done by the human worker
while the robot performs tasks 2 and 3. As consequence the tact time of the hybrid
system reduces to 100 s (instead of 140 s).
of huge and heavy
hile assembly tasks
more efficient with
s for a task differ if
ing or in a hybrid
imes of human and
ario human worker
io the task 1 can be
rforms tasks 2 and 3.
tem reduces to 100 s
se three alternatives
0 parts AB per year
à 8 h at 200 days per
or
d 1 robot.
cash inflows and
imated numbers of
he three alternative
t values (Table 3):
equentially
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 95
The tact time diagrams (Figs. 3.32a-3.32c) for these three alternatives look as
follows.
Fig. 40. Reduction of robot system costs compared to labour cost (above) and
assumed cost potentials of future hybrid automation (below) [92].
J. Krüger
et al. / CIRP
Manufacturing Technology 58 (2009) 62
J. Krüger et al. / CIRP Annals - Manufacturing
Technology
58 Annals
(2009) -628–646
done being.
by a robot
and
done by a robot and by a human
Furthermo
of regulations
human work
illness of human workers andillness
holiday
h
not taken
into
accoun
not taken into account. Nevertheless,
this
example
reduction
of the
t
high reduction of the tact timehigh
of a hybrid
system
com
manual solution
to
manual solution and to the automated
solution and
can b
for economic
advanta
for economic advantages of hybrid
solutions.
Lotter the
andperforma
Wienda
Lotter and Wiendahl [93] examined
assembly
systems,
wh
assembly systems, which include
automated
feedin
of parts.
By thi
sioning of parts. By this processsioning
time, which
is not
dir
the assembly
task ca
the assembly task can be reduced
significantly
an
the prod
efficiency of the production efficiency
process is of
increased.
assembly
wasfur
Fig. 41. Tact time diagram of human solution.
42. Tact
diagramsolution.
of robotic
solution.
hybrid
assembly was regarded hybrid
in comparison
with
solution.
(a) human solution.Fig. 42. Tact time diagram of robotic Fig.
(b)time
robotic
assembly
different
assembly for different products.
In six of for
seven
exam
costs per
piece
of a hy
costs per piece of a hybrid assembly
were
smaller
th
Table 4 shows as an example the calculation of the net present
assem
fully automated assembly. Thefully
cost automated
analysis given
in
value (NPV) of the hybrid solution, which is according to the NPV
shows
the for
electric
shows for the electric motors that
thefor
costs
a hyb
cellspieces
needed
anand
out
method the best solution.
cells needed for an output of 3000
perfor
day
needed
forlower
up tothan
6000t
This calculation is based on several assumptions, e.g. that all
needed for up to 6000 pieces per
day are
alternatives produce the same quality and that all tasks can be
automated assembly system. automated assembly s
Beside
other advan
Beside other advantages of hybrid
systems
are:
Human: all tasks sequentially
40
80
120
30
30
20
320
rep
hybrid systems represent a hybrid
form ofsystems
rationalisa
huma
integration of human and by integration
this giving of
positive
occupation,
1 and 2,3 parallel, rest sequentially
Best of
occupation,
assembly
costsare
perco
p
assembly costs per piece for hybrid
assembly
40
40
a wide
range
lot si
a wide range of lot sizes compared
with
fullyofautom
20
20
Fig. 43. Tact time diagram of hybrid
solution. costs per piece assembly
costs per
20
20
Fig. 43. Tact time diagram
of hybrid solution.
assembly
for
hybrid assembly
(c) hybrid
solution.
20
30
economic
even for r
economic even for relatively low
lot sizes,
20
30
use of hybrid
the use of hybrid systems isthe
characterized
by rs
20
20
Figura 3.32: Tact
Table
2time diagram.
investment,
which h
Table 2
investment, which helps to avoid
false investmen
Output
calculation
for human, robot and hybrid system assembly.
Output calculation for human, robot160
and
hybrid
system assembly.
100
the
low degree
of au
the low degree of automation
of hybrid
systems
in
reusefinishing
value forthe
a new
Human
reuse
value forHybrid
a new1product after
pro
Robot
Human
Hybrid 1 Robot
old one [93].
[93].
Total tact time [s/pc]
220.0 old one
320.0
100.0
Total tact time [s/pc]
220.0
320.0
Output
per
16.4
11.3 that 36.0
Assuming that the
output is
75,000
parts
AB100.0
per year
and
this
Outputneeded
per hour [pcs/h]
16.4hour [pcs/h]
11.3
36.0
Hours per8.0
shift [h/shift]8.0
8.0
8.0
8.0
Hours per shift [h/shift]
8.0
Further sources of
Further
sources
of
day [shifts/d]
3.0
3.0
3.0economic advantages of hyb
company is producing
inday2[shifts/d]
shifts á 8 hShifts
atper200
days3.0 per year,
the
company
will
Shifts per
3.0
3.0
be that aa better
hybridq
might
be
that
a
hybrid
solutionmight
will produce
Work days
per year [d/a]
250.0
250.0
250.0
Work days per year [d/a]
250.0
250.0
250.0
some
tasks cannot
in a given
Total 6,000.0
work time per6,000.0
year [h/a] 6,000.0 6,000.0
6,000.0in a given
6,000.0
some tasks
assembly
scenario
or c
Total work time per year [h/a]
need either (Table 3.2):
Output per year of 1 system [pcs/a]
98,181.8
67,500.0
216,000.0
high efforts be done w
Output per year of 1 system [pcs/a]
Requested output per year [pcs/a]
Needed robots (integer)
Needed humans per shift
Needed humans total
98,181.8
67,500.0
216,000.0
high efforts be done with a manual solution or with
Requested
output215,000.0
per year [pcs/a]
215,000.0
215,000.0
solution.
215,000.0
215,000.0215,000.0
solution. 0.0
Needed robots
(integer)
3.0
1.0
3.0
0.0
1.0
Based
experien
Based 4.0
on experience
achieved
withonintelligen
Needed humans
per
shift
0.0
1.0
0.0
4.0
1.0
devices
(IADs)
in asse
Needed humans
total 12.0
0.0
12.0 in assembly
3.0
devices (IADs)
tasks,
Stanley
Automatio
0.0
3.0
• 10 (9.6) workers (human solution) or
• 3 (2.2) robot systems (automated solution) or
• 1 hybrid system with 3 human workers and 1 robot.
reduction of
injury
reduction of injury costs, management
costs
andc
in of
total
which in total may reach anwhich
amount
58%may
thusr
payback
time
less t
payback time of less than 1 month
[95]
(Fig.of44).
45device
showswhic
an
Fig. 45 shows an intelligent Fig.
assist
Table 3
handling
of heavy load
Table 3
handling
of heavy loads. It is the
first admittance
disp
Cost
comparison
between
human,
robot
and
hybrid
system
assembly.
Cost comparison between human, robot and hybrid system assembly.
in industry.
to the
in industry. Due to the admittance
display Due
structure
the
apparent
Robot the apparent
Hybrid inertia ofsystem
system
the load
can
be reduci
Human
Robot
Hybrid Human
of 10 and
more,
so th
of 300,000.00
10 and more,
so that the
can
be perform
Invest
eel
0.00 ss
s325,000.00
s handling
Invest per robot eel
0.00
s per robot
300,000.00
s325,000.00
strain of the worker.
Invest per workplace
10,000.00
s0.00 s
15,000.00 s
strain
of the worker.
Invest per workplace
10,000.00
s0.00 s
15,000.00
s
Furthermore we assume the following cash inflows and outflows for the investments.
According to these
cash flows as the
estimated
numbers
of human
workers,
Variable
for robot cell
p.a.
0.00
5,000.00 s 5,000.00 s
Variable costs for robot cell p.a.
0.00
s costs5,000.00
s 5,000.00
s s
Variable
costs
for
human
worker
p.a.50,000.00
s0.00
s
55,000.00 s
Variable
costs the
for human
workeralternative
p.a.50,000.00 s0.00
s
55,000.00
s reach the following
workplaces and robot
cells
three
solutions
will
Revenues p.a.
750,000.00
s 750,000.00 s 750,000.00 s
Revenues p.a.
750,000.00
s 750,000.00 s 750,000.00
s
Interest rate 10%
10%
10%
10%
Interest rate
10%
10%
net present values (Table
3.3):
• manual solution: 760, 238.93 e
4
Table
• robotic solution:
2,4 117, 849.19 e Table
NPV calculation for hybrid system assembly.
NPV calculation for hybrid system assembly.
• hybrid solution:Hybrid
2, 754, 257.19
e
Hybrid
Outflow DCF
Inflow
Outflow
Inflow
Year
Year
Year
Year
Year
Year
Year
Year
Year
NPV
0
1
2
3
4
5
6
7
8
340,000.00
170,000.00
170,000.00
170,000.00
170,000.00
170,000.00
170,000.00
170,000.00
170,000.00
s
s
s
s
s
s
s
s
s
Year
0 s
0.00
Year
1
750,000.00
Year
2
750,000.00
Year
3
750,000.00
Year
4
750,000.00
Year
5
750,000.00
Year
6
750,000.00
Year
7
750,000.00
Year
8
750,000.00
NPV
340,000.00
s
340,000.00
s 0.00 s
s
s 170,000.00
527,272.73
s 750,000.00
170,000.00
s
s
479,338.84 s 750,000.00
s
s 170,000.00
435,762.58
s 750,000.00
s
s 170,000.00
396,147.80
s 750,000.00
s
s 170,000.00
360,134.37
s 750,000.00
s
s 170,000.00
327,394.88
s 750,000.00
s
s 170,000.00
297,631.71
s 750,000.00
s
s 170,000.00
270,574.28
s 750,000.00
DCF
s
s
s
s
s
s
s
s
340,000.00 s
527,272.73 s
479,338.84 s
435,762.58 s
396,147.80 s
360,134.37 s
327,394.88 s
297,631.71 s
270,574.28 s
2,754,257.19 s
Fig. 44.assist
Cost devices
reductio
Fig. 44. Cost reduction by use of intelligent
Table 3.4 shows as an example the calculation of the net present value (NPV)
of the hybrid solution, which is according to the NPV method the best solution.
2,754,257.19 s
96
CAPITOLO 3. THE ARTICLE SELECTED
Tabella 3.2: Output calculation for human, robot and hybrid system assembly.
Total tact time [s/pc]
Output per hour [pcs/h]
Hours per shift [h/shift]
Shifts per day [shifts/d]
Work days per year [d/a]
Total work time per year [h/a]
Output per year of 1 system [pcs/a]
Requested output per year [pcs/a]
Needed robots (integer)
Needed humans per shift
Needed humans total
Robot
Human
Hybrid 1
220.0
16.4
8.0
3.0
250.0
6,000.0
98,181.8
215,000.0
3.0
0.0
0.0
320.0
11.3
8.0
3.0
250.0
6,000.0
67,500.0
215,000.0
0.0
4.0
12.0
100.0
36.0
8.0
3.0
250.0
6,000.0
216,000.0
215,000.0
1.0
1.0
3.0
Tabella 3.3: Cost comparison between human, robot and hybrid system assembly.
Invest per robot eel
Invest per workplace
Variable costs for robot cell p.a.
Variable costs for human worker p.a.
Revenues p.a.
Interest rate
Robot
Human
Hybrid 1
0.00 e
−10, 000.00 e
0.00 e
−50, 000.00 e
750, 000.00 e
10%
−300, 000.00 e
0.00 e
−5, 000.00 e
0.00 e
750, 000.00 e
10%
−325, 000.00 e
−15, 000.00 e
−5, 000.00 e
−55, 000.00 e
750, 000.00 e
10%
Tabella 3.4: NPV calculation for hybrid system assembly.
Hybrid
Year 0
Year 1
Year 2
Year 3
Year 4
Year 5
Year 6
Year 7
Year 8
NPV
Outflow
Inflow
DCF
−340, 000.00 e
−170, 000.00 e
−170, 000.00 e
−170, 000.00 e
−170, 000.00 e
−170, 000.00 e
−170, 000.00 e
−170, 000.00 e
−170, 000.00 e
0.00 e
750, 000.00 e
750, 000.00 e
750, 000.00 e
750, 000.00 e
750, 000.00 e
750, 000.00 e
750, 000.00 e
750, 000.00 e
−340, 000.00 e
527, 272.73 e
479, 338.84 e
435, 762.58 e
396, 147.80 e
360, 134.37 e
327, 394.88 e
297, 631.71 e
270, 574.28 e
2, 754, 257.19 e
This calculation is based on several assumptions, e.g. that all alternatives produce the same quality and that all tasks can be done by a robot and by a human
being. Furthermore aspects of illness of human workers and holiday regulations
have not been not taken into account. Nevertheless, this example shows that a
high reduction of the tact time of a hybrid system compared to the manual solution and to the automated solution can be a good basis for economic advantages
of hybrid solutions.
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 97
Lotter and Wiendahl (2006) examined the performance of hybrid assembly
systems, which include automated feeding and provisioning of parts. By this
process time, which is not directly used for the assembly task can be reduced
significantly and the overall efficiency of the production process is increased.
This type of hybrid assembly was regarded in comparison with fully automated assembly for different products. In six of seven examined cases the costs per
piece of a hybrid assembly were smaller than those of a fully automated assembly. The cost analysis given in Lotter and Wiendahl (2006) shows for the electric
motors that the costs for a hybrid assembly cells needed for an output of 3000
pieces per day and also two cells needed for up to 6000 pieces per day are lower
than those of a fully automated assembly system.
Beside other advantages of hybrid systems are:
• hybrid systems represent a form of rationalisation with an integration of
human and by this giving positive impulses for occupation,
• assembly costs per piece for hybrid assembly are competitive for a wide
range of lot sizes compared with fully automated systems,
• assembly costs per piece for hybrid assembly systems are economic even
for relatively low lot sizes,
• the use of hybrid systems is characterized by relatively low investment,
which helps to avoid false investments,
• the low degree of automation of hybrid systems increases their reuse value
for a new product after finishing the production of an old one (Lotter and
Wiendahl, 2006).
Further sources of economic advantages of hybrid solutions might be that a
hybrid solution will produce a better quality or that some tasks in a given assembly scenario cannot or can only with high efforts be done with a manual solution
or with an automated solution.
Based on experience achieved with intelligent automation devices (IADs) in
assembly tasks, Stanley Automation estimates a reduction of injury costs, management costs and labour costs, which in total may reach an amount of 58% thus
leading to a payback time of less than 1 month.
(www.stanleyassembly.com/documents/en/VisteonCaseStudy.pdf.) (Fig. 3.33).
Fig. 3.34 shows an intelligent assist device which is used for handling of heavy
loads. It is the first admittance display prototype in industry. Due to the admittance display structure of the control system the apparent inertia of the load can
be reduced by a factor of 10 and more, so that the handling can be performed
with low strain of the worker.
, robot and hybrid system assembly.
Human
Robot
Hybrid
0.00 s
300,000.00 s325,000.00 s
10,000.00 s0.00 s
15,000.00 s
0.00 s
5,000.00 s 5,000.00 s
er p.a.50,000.00 s0.00 s
55,000.00 s
750,000.00 s 750,000.00 s 750,000.00 s
10%
10%
10%
98
which in total may reach an amount of 58% thus leading to a
payback time of less than 1 month [95] (Fig. 44).
Fig. 45 shows an intelligent assist device which is used for
handling of heavy loads. It is the first admittance display prototype
in industry. Due to the admittance display structure of the control
system the apparent inertia of the load can be reduced by a factor
of 10 and more, so that the handling can be performed with low
CAPITOLO 3. THE ARTICLE
strain of the worker.
SELECTED
m assembly.
Inflow
0.00 s
750,000.00
750,000.00
750,000.00
750,000.00
750,000.00
750,000.00
750,000.00
750,000.00
DCF
s
s
s
s
s
s
s
s
340,000.00 s
527,272.73 s
479,338.84 s
435,762.58 s
396,147.80 s
360,134.37 s
327,394.88 s
297,631.71 s
270,574.28 s
2,754,257.19 s
44. Cost reduction by use of intelligent assist devices (IAD) [95].
Figura 3.33:Fig.
Cost
reduction by use of intelligent assist devices (IAD).
644
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
production are normally handled within 20–
According to an unpublished internal analysi
a major German automotive company, seriou
by use of semi-automated flexible robotic
feeding of workpieces to the worker. By thi
could be reduced by about 30 min per car.
This can be achieved by a hybrid assem
handling for the final assembly step is fu
outstanding senso-motoric ability of the work
efficient support by intelligent assist syste
important advantage of IAS is the abdication
which normally is needed, when feeding an
pieces is done by robots close to the worker
In Europe robot based handling and
determine more than 65% of the robotic
application fields, where higher flexibility o
essentially improve the production proce
Fig. 45. First admittance display prototype in industry [3].
Figura 3.34: First admittance
display prototype in industry (Krüger JAmericas,
et al. 2006).
cleanroom applications for robo
whereas assembly was decreasing in 2007 [96
regarded as indicators for a future stronge
5.3. Markets for hybrid assembly systems
robot based assembly in Europe compar
Americas. The ongoing rapid changes in glob
In 2007
the worldwide
stocksystems
of industrial robots reached the
3.1.5.3 Markets for
hybrid
assembly
in the automotive sector will have a signifi
amount of about 1 million systems. The yearly supply climbed to
robot market in the future. The relation betwe
about 60,000 industrial robots in Asia/Australia, 30,000 in Europe
In 2007 the worldwide
industrial
the amount
of costs
about
production
which essentially determin
and aboutstock
20,000of
systems
in the robots
Americas. reached
While handling
automated or hybrid assembly systems will
operations still is the main field of application, other fields differ
1 million systems.
The yearly supply climbed to about 60,000 industrial
robots
by these dynamic changes.
in these regions. Although automotive industry is still the main
customer
significant
growth
can be in the Americas.
in Asia/Australia,
30,000forinindustrial
Europerobots,
and aabout
20,000
systems
observed in non-automotive sectors such as glass or food industry.
6. Conclusion and outlook
While handling operations
is the main
fieldsensing,
of application,
Beside others still
improvements
in force
environment other fields differ
recognition, human–machine-interfaces
and is
safety
tech- customer
The close
of human and
in these regions. Although
automotive industry
stillsystem
the main
forcooperation
innology are forecasted to be future technical trends [96]. These
assembly is motivated by the increased
dustrial robots, atrends
significant
growth
can
be observed
inoperation.
non-automotive
sectors
are significant
enablers
for hybrid
human–robot
adaptability
and reusability of assembly syste
An exemplary look to Europe shows the major market
systems can essentially reduce the amount
such as glass or food industry. Beside others improvements in force sensing, enconditions for hybrid assembly. For these systems the target
costs in relation to variable costs. The ef
vironment recognition,
and
safety
system
technology
industrieshuman-machine-interfaces
(automotive, household appliances and
aircraft)
face
systems
depends on the lot size, but also
strong global competition and represent a major part of the
cooperative workplace and its automated
are forecasted to European
be future
technical trends (World Robotics 2008).effectiveness
These trends
manufacturing industry. Any effort to improve the
of hybrid assembly also depen
competitiveness,
productivity
and
market
responsiveness
of
feeding
of workpieces to the cooperative wo
are significant enablers for hybrid human-robot operation.
these industries would be extremely important for Europe. The
An essential precondition for the time effi
An exemplaryautomotive
look to industry
Europe
shows
the major
conditions
for hybrid
employs
2.2 million
people market
in production
in
tion between
human and machine is the safe
nearly all member states of the EU. There are 380,000 SME’s in the
of research has been carried out in this field in
assembly. For these
systems the target industries (automotive, household
apsupply chain for the automotive industry in Europe. In 2002 global
sensor systems for the surveillance of the
production
passengerglobal
cars and competition
light and heavy trucks
just
human
and robots
pliances and aircraft)
faceofstrong
andwas
represent
a major
partare available on the mark
over 59 million units. The main vehicle-producing areas are the
The research in the field of intelligent ass
Asia-Pacific (19.3 million), western Europe (17.4 million) and
efficiently support the worker is the basis to k
North America (16.8 million). The turnover of European carloop in order to utilize his cognitive and senso
makers over the last few years has been on average 340 billion
for highly flexible assembly. The suppor
Euros per year. The European industry of household appliances
comprise safety as well as reduction of physic
employs 200,000 people directly and represents a yearly turnover
synchronisation between worker and autom
of about 35 billion Euros. 40,000 SME’s are suppliers for that
Future research will have to focus on differ
industry. Approximately 50 million large appliances (washing
cooperative hybrid assembly. The robustnes
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES 99
of the European manufacturing industry. Any effort to improve the competitiveness, productivity and market responsiveness of these industries would be
extremely important for Europe. The automotive industry employs 2.2 million
people in production in nearly all member states of the EU. There are 380,000
SME’s in the supply chain for the automotive industry in Europe. In 2002 global
production of passenger cars and light and heavy trucks was just over 59 million units. The main vehicle-producing areas are the Asia-Pacific (19.3 million),
western Europe (17.4 million) and North America (16.8 million). The turnover
of European carmakers over the last few years has been on average 340 billion
Euros per year. The European industry of household appliances employs 200,000
people directly and represents a yearly turnover of about 35 billion Euros. 40,000
SME’s are suppliers for that industry. Approximately 50 million large appliances
(washing machines, refrigerators, etc.) and 200 million small ones are produced
in Europe per year. The aerospace manufacturing industry had in Europe a
turnover of 81 billion Euros in 2003. Over 435,000 people are directly employed in
high-quality jobs. Its business has a pervasive influence up and down the supply
chain, beyond the sector to some 80,000 SME’s within the European Union.
3.1.5.4
Hybrid assembly in automotive industry
In automotive industry the degree of automation in assembly mainly depends on
two aspects: costs and complexity.
Power train assembly in western European countries is fully automated. Manual operation is typical for assembly of flexible interior parts of the car. In this
area, automated handling with a robot is difficult due a lack of automated gripping flexibility and sensitivity and also because of danger of surface damages of
the component. Manual or hybrid assembly tasks in automotive production are
normally handled within 20-30 assembly stations. According to an unpublished
internal analysis of assembly tasks of a major German automotive company, serious savings can be made by use of semi-automated flexible robotic systems for
faster feeding of workpieces to the worker. By this, manual work load could be
reduced by about 30 min per car.
This can be achieved by a hybrid assembly structure where handling for the
final assembly step is fulfilled by use of the outstanding senso-motoric ability of
the worker in combination by efficient support by intelligent assist systems (IAS)
(World Robotics 2008, Bernhardt R et al. 2007). An important advantage of IAS
is the abdication of safety equipment, which normally is needed, when feeding
and handling of workpieces is done by robots close to the worker.
In Europe robot based handling and assembly operations determine more
than 65% of the robotic market. These are application fields, where higher flexibility of hybrid systems can essentially improve the production process. In Asia
and the Americas, cleanroom applications for robotics were increasing whereas
100
CAPITOLO 3. THE ARTICLE SELECTED
assembly was decreasing in 2007 (World Robotics 2008). These facts may be regarded as indicators for a future stronger diffusion of hybrid robot based assembly in Europe compared to Asia and the Americas. The ongoing rapid changes
in global markets especially in the automotive sector will have a significant influence to the robot market in the future. The relation between fixed and variable
production costs which essentially determines the use of purely automated or
hybrid assembly systems will certainly be affected by these dynamic changes.
3.1.6
Conclusion and outlook
The close cooperation of human and machine in hybrid assembly is motivated
by the increased need for flexibility, adaptability and reusability of assembly systems. Hybrid assembly systems can essentially reduce the amount of fixed production costs in relation to variable costs. The effectiveness of these systems depends on the lot size, but also on the design of the cooperative workplace and its
automated systems. The overall effectiveness of hybrid assembly also depends
on the intelligent feeding of workpieces to the cooperative workplace.
An essential precondition for the time efficient close cooperation between human and machine is the safety of the worker. A lot of research has been carried
out in this field in recent years and first sensor systems for the surveillance of the
interaction between human and robots are available on the market.
The research in the field of intelligent assist devices (IAD) that efficiently support the worker is the basis to keep the worker in the loop in order to utilize his
cognitive and senso-motoric advantages for highly flexible assembly. The support functions of IADs comprise safety as well as reduction of physical strain
and efficient synchronisation between worker and automated systems.
Future research will have to focus on different aspects for highly cooperative
hybrid assembly. The robustness of the system parts for safety, load reduction
and others will have to be increased in order to extend the field of applications. A
new challenge will be to improve the cooperation of human and machine not only
for a single human but also for a group of workers that have to fulfil a common
task supported by one or more machines at the same time (Fig. 3.35).
The close cooperation between human and machine may in the future also
change the way that automated systems will be programmed. It can be foreseen,
that a wide range of automated cinematic systems such as robots will in the future include interfaces for efficient teach modes based on the direct force coupled
interaction between human and robot.
The human oriented automation approach may generally lead to a change in
the design principles for robots. Today design of industrial robots is focused for
accuracy. Control with application of external sensors is needed for stable and
safe interaction. In long term solutions the design of intrinsically safe robots will
focus on safety and sophisticated control will provide accuracy.
3.1. COOPERATION OF HUMAN AND MACHINES IN ASSEMBLY LINES101
J. Krüger et al. / CIRP Annals - Manufacturing Technology 58 (2009) 628–646
[17]
[18]
[19]
[20]
[21]
[22]
[23]
Fig. 46. Application of cobotic devices in future human–robot–human interaction
[24]
Interactive Communication, ROMAN2002, Berlin, Ge
399–404.
Schraft RD, Helms E, Hans M, Thiemermann S (200
tion and Co-operation for Mobile and Assisting Rob
2004, Madeira, 67–77.
In: www.smerobot.org.
Iossifidis I, Bruckhoff C, Theis C, et al, (2002) Cora: A
Assistant for Human Environment. Proceedings of th
on Robot and Human interactive Communication, ROM
September 25–27, 392–398.
Akella M, Peshkin M, Colgate E, Wannasuphoprasit
Material Handling Technology. Proceedings of the AS
1–7.
Kazerooni H (1995) The Human Power Amplifier Te
of California, Berkeley DSC Vol. 57-2. IMECE Proceed
Systems and Control Division, ASME, 605–613.
Delnondedieu Y, Trocazz J (1995) PADyC a Passive
straint: A Prototype with Two DOF. Proceedings
Medical Robotics and Computer Assisted Surgery, 173
Surdilovic D, Radojicic J (2007) Robust Control o
Interfaces. Proceedings of the ICRA, Rome, 3237–32
Surdilovic D, Bernhardt R (2005) Novel Interactive
ings of the 16th IFAC World Congress, Prague, Czech R
full paper nr. 05173).
Colgate JE (1988) The Control of Dynamically Intera
Massachusetts Institute of Technology.
Colgate JE, Hogan N (1988) Robust Control of Dyn
tems. International Journal of Control 48(1):65–88.
Dohring M, Newman W (2003) The passivity of n
implementations, Robotics and Automation, 2003.
International Conference on Robotics and Automation
3710–3715.
Adams RJ, Hannaford B (1998) A Two-port Fram
Unconditionally Stable Haptic Interfaces. Int’l Conf
Systems (Victoria, BC), August, 1254–1259.
Adams RJ, Hannaford B (1999) Stable Haptic Inter
onments. IEEE Transactions on Robotics and Automat
Buerger SP, Hogan N (2006) Relaxing Passivity for H
Proceedings of the Int. Conference on Intelligent Ro
4570–4575.
Sugano S (2000) A Robotic Co-operation System Ba
Approached Human Work Model. Proceedings of the
Automation, ICRA, 4058–4063.
Colgate JE, Hogan N (1989) An Analysis of Conta
Passive Physical Equivalents. Proceedings of the IEEE
and Automation, 404–409.
Vukobratović M, Surdilović D (2002) Control of Ro
Tasks. in Nwokah OI, Hurmuzlu Y, (Eds.) ‘The Mecha
book’: Modeling, Measurement and Control. CRC Press
Book W, Charles H, Davis H, Gomes M (1996) The Con
of a Passive Trajectory Enhancing Robot. Proceedi
Systems Control Division, DSC-Vol. 58, 633–638.
Stopp A, Horstmann S, Kristensen S, Lohnert F (2
Learning for Manufacturing Assistants. IEEE Transac
nics 50(August (4)):705–707.
Bernhardt R, Surdilovic D, Katschinski V, Schreck
Generation of Flexible Assembly Systems. 8th In
Information Technology for Balanced Automation S
Portugal, June 23–25, .
Bernhardt R, Surdilovic D, Katschinski V, Schröer K
Systems through Human Integration. Proceedings of
Conference on Distributed Human–Machine Systems,
12, 497–502.
Albu-Schäffer A, Haddadin S, Ott Ch, Stemmer A,
(2007) Industrial Robot 34(5):376–385.
Brecher C, Schröter B, Almeida C (2005) Developm
Portable Robot Systems for Material Handling Tas
International Conference on Reconfigurable Manufac
gan, USA), May 10–12, 1–8.
Som F (2006) Safe Robot Control for a Personne
Separating Safety Devices. Proceedings of the Robo
Berichte 1841, 745–752.
Zinn M, Khatib O, Roth B (2004) A New Actuati
Friendly Robot Design. Proceedings of the ICRA’04, IEE
on Robotics and Automation, vol. 1, 249–254.
Oberer S, Schraft R-D (2007) Robot-Dummy Cras
Assessment. IEEE International Conference on Robot
Italy, April 10–14, 2934–2939.
Kroth E (2007) Flexible Robot-based Automization
Proceedings of the PTK2007, Berlin, Germany, Octob
In: www.industrialnetworking.co.uk/mag/v9-1/f_safety.htm
In: www.sick.de.
In: www.pilz.de/products/sensors/camera/f/safetyeye/index
Kochan A (2006) Robots and Operators Work Hand i
International Journal 33(6):422–424.
Krüger J, Nickolay B, Heyer P, Seliger G (2005) Image
Flexible Man–Robot-Cooperation. Annals of the CIR
Winkler B (2007) Safe Space Sharing Human–Robo
Time-of-Flight Camera. Robotic Industries Associati
Vision Show: Technical Conference Proceedings, Ros
June 12–14, 1–8.
Figura 3.35: Application
ofassembly
cobotic
for airplane
[97].devices in future human-robot-human interaction for
airplane assembly (Krüger J et al. 2008).
[25]
By this, the next generation of robots will interact with human
[26]
directly for cooperative manipulation, where the robot is
[27]
responsible for load-bearing and precision whereas the human
By this, the next
generation
of robots
willand
interact
with human directly for
can contribute
with sensing,
intelligence
skills.
Generally the
advantages
of a close
cooperation between
cooperative manipulation,
where
the robot
is responsible
for load-bearing
and
[28]
human and intelligent assist systems in assembly lines build a
precision whereasbasis
theforhuman
canand
contribute
with sensing,
intelligence and skills.
a more flexible
also more sustainable
way of assembly
[29]
and disassembly in the future.
Generally the advantages of a close cooperation between human and intelli[30]
gent assist systems
in assembly lines build a basis for a more flexible
and also
Acknowledgements
more sustainable way
of assembly and disassembly in the future. [31]
The authors would like to thank Prof. M. Leu, Prof. G. Seliger,
Prof. S. Sugano, Prof. S. Takata, Prof. H.-P. Wiendahl and all other
colleagues who sent valuable contributions to prepare this article.
[32]
References
[33]
[1] Bley H, Reinhart G, Seliger G, Bernardi M, Korne T (2004) Appropriate Human
Involvement in Assembly and Disassembly. CIRP Annals – Manufacturing
Technology 53(2):487–509.
[2] www.assistor.de.
[3] Krüger J, Bernhardt R, Surdilovic D, Seliger G (2006) Intelligent Assist Systems
for Flexible Assembly. Annals of the CIRP 55:29–33.
[4] Lien TK, Rasch FO (2001) Hybrid Automatic-manual Assembly Systems. CIRP
Annals – Manufacturing Technology 50(1):21–24.
[5] Zaeh MF, Prasch M (2007) Systematic Workplace and Assembly Redesign for
Aging Workforces. Production Engineering Research and Development 1:57–64.
[6] Occupational Safety and Health Administration www.osha.eu.int.
[7] N.N., Intelligent Assist Devices, Stanley Cobotics, Brochure, www.stanleya
ssembly.com, 2005.
[8] Colgate JE, Wannasuphoprasit W, Peshkin M (1996) Cobots Robots for Collaboration with Human Operator. Proceedings of the ASME Dynamic Systems and
Control Division, DSC-Vol. 58, 433–440.
[9] Thiemermann S, Schraft RD (2003) team@work—Mensch-Roboter-Kooperation in der Montage. Automatisierungstechnische Praxis atp Praxis der MessSteuerungs- Regelungs- und Informationstechnik, vol. 45(11). . pp. S. 31–S.35.
[10] Schraft RD, Meyer C, Parlitz C, Helms E (2005) PowerMate—A Safe and Intuitive
Robot Assistant for Handling and Assembly Tasks. Proceedings of the 2005 IEEE
International Conference on Robotics and Automation, ICRA, April 18–22, 4074–
4079.
[11] Bernhardt R, Surdilovic D, Katschinski V, Schröer K (2007) Flexible Assembly
Systems through Workplace-Sharing and Time-Sharing Human–Machine
Cooperation. IFAC Workshop on Intelligent Manufacturing Systems (IMS’07),
Alicante Spain, May 23–25, 282–286.
[12] Asfour T, Berns K, Dillmann R (2000) The Humanoid Robot ARMAR: Design and
Control. International Conference on Humanoid Robots (Humanoids 2000), Boston (MIT), USA, September 7–8, 7–8.
[13] Ehrenmann M, Lütticke T, Dillmann R (2001) Dynamic Gestures as an Input
Device for Directing a Mobile Platform. International Conference on Robotics and
Automation (ICRA), Seoul, Korea, May 23–25, 21–26.
[14] Ehrenmann M, Becher R, Giesler B, et al, (2002) Interaction with Robot
Assistants: Commanding ALBERT. Proceedings of the 2002 IEEE/RSJ International
Conference on Intelligent Robots and Systems (IROS), Lausanne, Schweiz, September, 1–6.
[15] Theis C, Iossifidis I, Steinhage A (2001) Image Processing Methods for Interactive Robot Control. Proceedings of the 10th IEEE Inter. Workshop on Robot–
Human Interactive Communication (ROMAN’01), Paris, France, September 18–
21, 424–429.
[16] Helms E, Schraft RD, Hägele M (2002) rob@work: Robot Assistant in Industrial
Environments. Proceedings of the 11th IEEE Int. Workshop on Robot and Human
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
102
3.2
CAPITOLO 3. THE ARTICLE SELECTED
A new cell production assembly system with
human-robot cooperation
In manufacturing, where the wages of operators are high and the availability of
expert operators remains limited, it is proving impossible to improve the efficiency of cell production assembly systems. Even novice operators are asked to
achieve high levels of productivity and reliability with a diverse range of products. To satisfy this demand, the authors have developed a new cell production
assembly system with human-robot cooperation. This system consists of three
key technologies; parts feeding by double manipulators on a mobile base, production process information support for the operator, and safety management
for cooperation between the operator and the robot.
3.2.1
Introduction
Cell production assembly systems, in which human operators manually assemble products from start to finish, play an important role in manufacturing. Several cell production assembly systems have been proposed to improve productivity and enable application to a diverse range of products with quantitative
scalability.
Seki (2003) developed a production cell called the “Digital Yatai” which monitors the assembly progress and presents information about the next step in the
process. Reinhart and Patron (2003) developed an augmented reality system using a semi-transparent head mount display, through which information is provided to the human operator. These studies support the operator only from
information aspect, however.
Vasilash (1999) created a production cell called the “Raven Super Cell” which
consists of a horseshoe-shaped fixed table and a round movable worktable. To
reduce the operator’s physical burden and improve the assembly precision, Hayakawa et al. (1998) employed a manipulator to grasp the component parts during the assembly process. These improved the assembly cell only in the physical
support aspect.
The “Attentive Work Bench” by Sugi et al. (2005) aimed to support human
operators from both an information and physical aspect. In this system, a projector is employed to provide assembly information to the operator, and self-moving
trays are used to deliver parts to the operator. These devices are, however, only
experimental and cannot be applied to practical manufacturing.
To maintain the competitiveness of manufacturing in those countries where
wages are high, it would be advantageous if improved efficiency and cost reduction could be achieved by means of automation centered on industrial robots,
with additional tasks such as parts feeding being performed by human opera-
3.2. A NEW CELL PRODUCTION ASSEMBLY SYSTEM WITH
HUMAN-ROBOT COOPERATION
103
tors. Those same human operators would also take on any complicated tasks
that cannot be performed automatically by existing robots.
M. Morioka, S. Sakakibara / CIRP Annals - Manufacturing Technology 59 (2010
10
In the above-mentioned system, human-robot cooperation would be so inis more suitable for a cell
evitable that a safety management technology would have to be incorporated. In of modifying the system
teaching of the robot.
addition, as the present society is aging, with fewer young people, it is becoming
The operation suppor
tages;
extremely difficult to find a sufficient number of experienced operators (experts).
Full automation of par
Because of this, there is a need for an information support technology that would manipulators on a mob
Instruction of the asse
allow even inexperienced operators (novices) to quickly acquire the necessary between the manipulat
skills and perform work of high quality.
In general cell produc
from making a mistake in
The purposes of this study were as follows;
that includes all the parts
1. To develop a new cell production assembly system
which will realize greater
Fig. 1. Setup of new cell production.
productivity and higher quality of manufacturing than is possible with conshows the entire setup of the new cell production system, while
Fig. 2 shows
the system configuration.
ventional cell production. Achieve cost
competitiveness
in manufacturing
As shown in Fig. 2, in (A), the parts feeding station, based on
that can produce a diverse range of products
with
quantitative
scalability.
operation support
technology,
a robot automatically
feeds all the
parts necessary for assembly to the assembly station. In (B), the
2. To develop a new operation supportassembly
technology
foronsupporting
thetechnology,
human
station, based
information support
a
human operator can concentrate on complicated work, assisted by
operators from a physical and information
standpoint,
centering
mainly
by
both physical
support from the robots
by, for example,
clamping
parts as a fixture, and information support from the system. (C)
robots.
Safety management technology oversees the safe interaction
3.2.2
between stations (A) and (B), which ensures safe cooperation
between the human operator and the robot. The product being
assembled in this study was a cable harness with several
connecters and fastener plates. The assembly takes approx.
10 min when done by an experienced operator.
System configuration of a new cell production assembly
system
2.1. Operation support technology
One of advantages
of cell production
assembly is in
the which
ability to
In this study, the authors set out to design a new
cell production
system,
modify the system layout more flexibly than would be possible with
automatic
conveyer
in order
to be able to
robots and human operators take on thosea full
tasks
that
arelinebest
suited
toproduce
theira diverse
parrange of products with quantitative scalability. In this study, to
the modification
of the system
layout,
authors
developed
ticular characteristics. The new system hasenable
three
subsystems;
(A)
A the
parts
feeding
a robot with double manipulators of a size equivalent to human arms
station, (B) An assembly station and (C)forSafety
technology.
Fig.
high-speedmanagement
bin picking, mounted on
a mobile base capable
of
moving in all directions and spinning, as shown in Fig. 3(a).
3.36a shows the entire setup of the new cellThis
production
while
Fig. base
3.36b
robot is driven by system,
a battery installed
in the mobile
and
is equipped with wireless communication systems and vision
shows the system configuration.
systems to compensate for any errors in the position and posture of
the mobile base. In comparison with a traditional manipulator that
runs on rails (typically installed in machining factories), this robot
10
M. Morioka, S. Sakakibara / CIRP Annals - Manufacturing Technology 59 (2010) 9–12
is more suitable for a cell production assembly from the viewpoint
of modifying the system layout easily, simply by modifying the
teaching of the robot.
The operation support technology has the following advantages;
Full automation of parts kitting and feeding by using double
manipulators on a mobile base.
Instruction of the assembly target position by collaboration
between the manipulator and a laser pointer during assembly.
Fig. 1. Setup of new cell production.
(a) Setup.
shows the entire setup of the new cell production system, while
Fig. 2 shows the system configuration.
As shown in Fig. 2, in (A), the parts feeding station, based on
operation support technology, a robot automatically feeds all the
parts necessary for assembly to the assembly station. In (B), the
assembly station, based on information support technology, a
human operator can concentrate on complicated work, assisted by
both physical support from the robots by, for example, clamping
parts as a fixture, and information support from the system. (C)
Safety management technology oversees the safe interaction
between stations (A) and (B), which ensures safe cooperation
between the human operator and the robot. The product being
assembled in this study was a cable harness with several
In general cell production assembly, an operator is prevented
from making a mistake in assembly by the preparation of a parts kit
that includes all the parts necessary for assembly, which is fed to
the operator. In this study, as shown in Fig. 3(b), the use of simple
additional devices and the mobile double manipulators, equipped
with vision systems on their arms, enables the automation of parts
Fig.the
2. System
configuration
of new
production.
feed, including
extraction
of a parts
bin cell
from
the parts shelves,
(b)
System
configuration.
the picking of parts from that bin, kitting of parts onto a tray,
visually checking the parts in the tray, and then feeding the parts
kit to the operator. This robot offers cost reduction and highreliability of parts kitting thanks to the simple additional
surrounding devices such as the common part shelves and visual
checking. In addition, since LED lights are installed in the vision
systems, the manipulators can kit parts with high reliability,
regardless of the ambient light level.
In assembling the cable harness, the authors demonstrated bin
picking of nine different parts, making two kinds of parts kits and
feeding them to the operator by the mobile manipulators. With
this demonstration, the authors were able to confirm that this
parts feeding automation system can run continuously and with
high reliability.
Figura 3.36: New cell production.
the operator. In this study
additional devices and th
with vision systems on th
feed, including the extrac
the picking of parts from
visually checking the par
kit to the operator. This
reliability of parts kitt
surrounding devices such
checking. In addition, sin
systems, the manipulato
regardless of the ambien
In assembling the cabl
picking of nine different p
feeding them to the ope
this demonstration, the
parts feeding automation
high reliability.
This physical suppo
productivity by releasin
obtaining parts and inste
cated assembly tasks.
On the other hand,
assembles the parts one
assembly and reduce his
ulators can assist the op
during assembly, as show
target positions are instr
between the manipulato
assembly station, even
without making mistakes
2.2. Information support t
In this study, the autho
Support System (MASS) th
master frequently changi
and then go on to execute
shows the software stru
advantages;
Detailed comprehensib
multi-modal informatio
an assembly table with
The amount and degre
controlled according to
This system consists
TransfER (MASTER) and
SupportER (MAISER), as s
MASTER mainly utiliz
tracking using IP camera
motions of assembly expe
them.
MAISER provides de
novice operators by disp
104
CAPITOLO 3. THE ARTICLE SELECTED
As shown in Fig. 3.36b, in (A), the parts feeding station, based on operation
support technology, a robot automatically feeds all the parts necessary for assembly to the assembly station. In (B), the assembly station, based on information
support technology, a human operator can concentrate on complicated work, assisted by both physical support from the robots by, for example, clamping parts
as a fixture, and information support from the system. (C) Safety management
technology oversees the safe interaction between stations (A) and (B), which ensures safe cooperation between the human operator and the robot. The product
being assembled in this study was a cable harness with several connecters and
fastener plates. The assembly takes approx. 10 min when done by an experienced
operator.
3.2.2.1
Operation support technology
One of advantages of cell production assembly is the ability to modify the system
layout more flexibly than would be possible with a full automatic conveyer line
in order to be able to produce a diverse range of products with quantitative scalability. In this study, to enable the modification of the system layout, the authors
developed a robot with double manipulators of a size equivalent to human arms
for high-speed bin picking, mounted on a mobile base capable of moving in all
directions and spinning, as shown in Fig. 3.37a.
(a) Mobile manipulators.
(b) Parts kitting.
Figura 3.37: Mobile manipulators and parts kitting.
This robot is driven by a battery installed in the mobile base and is equipped
with wireless communication systems and vision systems to compensate for any
3.2. A NEW CELL PRODUCTION ASSEMBLY SYSTEM WITH
HUMAN-ROBOT COOPERATION
105
errors in the position and posture of the mobile base. In comparison with a traditional manipulator that runs on rails (typically installed in machining factories),
this robot is more suitable for a cell production assembly from the viewpoint
of modifying the system layout easily, simply by modifying the teaching of the
robot.
The operation support technology has the following advantages;
• Full automation of parts kitting and feeding by using double manipulators
on a mobile base.
• Instruction of the assembly target position by collaboration between the
manipulator and a laser pointer during assembly.
In general cell production assembly, an operator is prevented from making a
mistake in assembly by the preparation of a parts kit that includes all the parts
necessary for assembly, which is fed to the operator. In this study, as shown in Fig.
3.37b, the use of simple additional devices and the mobile double manipulators,
equipped with vision systems on their arms, enables the automation of parts feed,
including the extraction of a parts bin from the parts shelves, the picking of parts
from that bin, kitting of parts onto a tray, visually checking the parts in the tray,
and then feeding the parts kit to the operator. This robot offers cost reduction
and high reliability of parts kitting thanks to the simple additional surrounding
devices such as the common part shelves and visual checking. In addition, since
LED lights are installed in the vision systems, the manipulators can kit parts with
high reliability, regardless of the ambient light level.
In assembling the cable harness, the authors demonstrated bin picking of nine
different parts, making two kinds of parts kits and feeding them to the operator
by the mobile manipulators. With this demonstration, the authors were able to
confirm that this parts feeding automation system can run continuously and with
high reliability.
This physical support can produce an improvement in productivity by releasing human operators from the task of obtaining parts and instead letting them
concentrate on complicated assembly tasks.
On the other hand, in the assembly station, the operator assembles the parts
one by one. To increase the productivity of assembly and reduce his/her burden, one of the mobile manipulators can assist the operator by clamping a part
as a fixture during assembly, as shown in Fig. 3.38. In addition, since the assembly target positions are instructed step by step by the collaboration between the
manipulator and the laser pointer installed in the assembly station, even novice
operators can assemble parts without making mistakes.
2.3. Safety management technology
106
As mentioned in Section 2.1, in the assembly station, the highspeed, high-power manipulators, used for feeding parts, share the
workspace with the human operator and cooperate with him/her
CAPITOLO
3.the
THE
ARTICLE
in automatic operation. This situation
demanded
development
of a new method of safety management. Recently, various methods
reliability CPUs.
As shown in Fig. 6, when the hum
separated from that of the manipula
separate workspaces by a light curta
of cooperation between the human
as in the case of automatic operati
SELECTED
between the human operator and th
is of no use and the speed and motio
restricted. The speed restriction pre
being hit hard by the manipulators, a
prevents him/her from becoming t
tors and the surrounding devices
restrictions are overseen by two i
controller, high reliability is possibl
Fig. 3. Mobile manipulators and parts kitting.
of safety management to be applied to the cooperation between
tures, animation and voice
human operators and manipulators have been studied [7], with
with a built-in LCD monitor,
some of them utilizing laser scanner technology or camera systems
ors to read the assembly
for 3D workspace surveillance [8]. However, these safety manageoving his/her line of sight
ment systems are designed to stop manipulators upon detecting Fig. 5. Software structure of skill transfer sys
assembly table can improve
system.
Fig. 4. Assembly operation with assistant of manipulator.
the
approach
of human operation
operators and
to secure
the safety of
ors.
Figura
3.38: Assembly
withnot
assistant
of manipulator.
the manipulators as in automatic operation in cooperation with
is installed to enable the
human operators.
vidual operators, and given
In this study, the authors developed the following new method
provided can be controlled
support technology
of safety management
for manipulators, to be applied to automatic
or, this information3.2.2.2
support Information
operation in cooperation with human operators; restriction of
ovices to experts.
this study,
the authors
developed
the as
Multi-modal
Support System
motion
speed and
motion range,
checked by a Assembly
software system
wed even novices toIn
execute
the robot
controller,
configured
with two independent
high- changing
ctivity and reliability.
(MASS) thatinwas
intended
to help
novice operators
master frequently
reliability
CPUs.
assembly operations in a short time, and then go on to execute the operations
As shown in Fig. 6, when the human operator’s workspace can be
with high reliability.
Fig. 3.39 shows the software structure. This system has the
separated from that of the manipulator, the whole cell is divided into
following
advantages;
separate workspaces by a light curtain. On the other hand, in the case
assembly station, the
highof cooperation between the human operator and the manipulators,
for feeding parts, share the
as in the case of automatic operation with the workspace shared
nd cooperate with him/her
• Detailed
comprehensible
information
can be provided
including
between
the human operator
and the manipulators,
the light
curtain multi-modal
demanded the development
is of no use
and
the speedanimation
and motion range
of the manipulators
are table with
t. Recently, various methods
information
with
pictures,
and voice
by an assembly
restricted.
The
speed
restriction
prevents
the
human
operator
from
a horizontally built-in LCD monitor.
being hit hard by the manipulators, and the motion range restriction
• The amount
degree
detail oftrapped
the information
be controlled acpreventsand
him/her
fromofbecoming
between the can
manipulacording
to and
the skill
level of thedevices.
operator.
tors
the surrounding
Since the above-mentioned
restrictions are overseen by two independent CPUs in the robot
controller, high reliability is possible.
istant of manipulator.
Fig. 5. Software structure of skill transfer system and assembly information support
Figura 3.39: Software
structure of skill transfer system and assembly information support
system.
system.
3.2. A NEW CELL PRODUCTION ASSEMBLY SYSTEM WITH
HUMAN-ROBOT COOPERATION
107
This system consists of the Multi-modal Assembly Skill TransfER (MASTER)
and Multi-modal Assembly Information SupportER (MAISER), as shown in Fig.
3.39.
MASTER mainly utilizes motion capture based on visual mark tracking using
IP cameras, and evaluates the correct posture and motions of assembly experts to
extract the skill information from them.
MAISER provides detailed comprehensible instructions to novice operators
by displaying multi-modal information about assembly operations such as text,
pictures, animation and voice (Tan JTC, Duan F, et al, 2008). A new horizontal
assembly table with a built-in LCD monitor, as shown in Fig. 3.38, enables operators to read the assembly instructions, step by step, without moving his/her
line of sight away from the work in hand. This new assembly table can improve
productivity and reduce assembly errors.
In addition, since an ID system is installed to enable the management of the
skill level of individual operators, and given that the amount of information being provided can be controlled according to the skill level of the operator, this
information support system can be applied widely from novices to experts.
The development of MASS has allowed even novices to execute assembly
with a high degree of productivity and reliability.
3.2.2.3
Safety management technology
As mentioned in Section 3.2.2.1, in the assembly station, the high speed, highpower manipulators, used for feeding parts, share the workspace with the human
operator and cooperate with him/her in automatic operation. This situation demanded the development of a new method of safety management. Recently, various methods of safety management to be applied to the cooperation between human operators and manipulators have been studied (J. Krüger, T.K. Lien, A. Verl,
2009), with some of them utilizing laser scanner technology or camera systems
for 3D workspace surveillance (J. Krüger, B. Nickolay, P. Seliger, 2005). However, these safety management systems are designed to stop manipulators upon
detecting the approach of human operators and not to secure the safety of the
manipulators as in automatic operation in cooperation with human operators.
In this study, the authors developed the following new method of safety management for manipulators, to be applied to automatic operation in cooperation
with human operators; restriction of motion speed and motion range, as checked
by a software system in the robot controller, configured with two independent
high reliability CPUs.
As shownin Fig. 3.40, when the human operator’s workspace can be separated from that of the manipulator, the whole cell is divided into separateworkspaces
by a light curtain.Onthe other hand, in the case of cooperation between the human operator and the manipulators, as in the case of automatic operation with
108
CAPITOLO 3. THE ARTICLE SELECTED
the workspace shared between the human operator and the manipulators, the
light curtain is of no use and the speed and motion range of the manipulators are
restricted. The speed restriction prevents the human operator from being hit hard
by the manipulators, and the motion range restriction prevents him/her from becoming trapped between the manipulators and the surrounding devices. Since
the above-mentioned restrictions are overseen by two independent CPUs in the
robot controller,
high reliability is possible.M. Morioka, S. Sakakibara / CIRP Annals - Manufacturing Technology 59 (20
12
doubled. Note that the a
a minimum in all cases.
relatively easy and that
words, this system ma
product being manufac
In term of assembly
(insertion error) was ob
I), while in the new ce
prevented thanks to the
pointer guidance and th
According to the ex
authors can conclude
system can accelerate t
Fig.
6. Safety
management in
in the
of cooperation
between
operator operator
and
Figura 3.40:
Safety
management
thecase
case
of cooperation
between
and simultaneously p
while
manipulators.
manipulators.
4. Conclusions and fu
This new safety management enables the same manipulators to both kit parts
In this study, the a
at high speed in the parts feeding station and to assist in the assembly performed
assembly system, in wh
by the human operators, as in the case of automatic operation in cooperationprovided
with
to the human
operators in the assembly station. If the manipulators are to shift from parts with
feed-the robot while pro
performed. The safety
ing and instead assist in the assembly being performed by the human operators
from the viewpoint th
according to the progress of the assembly, higher productivity can be achieved.
human operators in aut
To evaluate the safety management of the entire system, including the abovecan achieve higher pro
mentioned method, the authors listed up 168 risks including reasonably foreseefor the assembly of a ca
can be doubled relative
able misuses, and set up countermeasures according to the “Three Step Method”
(inherently safe design measure, safeguarding or complementary protective tion.
mea-Furthermore, the
with the new cell produ
Fig. 7. Difference in assembly time for experts and novices.
sure, and information for use), as defined in ISO 12100 (2003) as a risk assessment
because assembly error
method. Since
risks have
yet to be
solved,
authors
will develop
funcThissome
new safety
management
enables
thethe
same
manipulators
to
In addition, the new
tional safety
including
the
monitoring
ofparts
the operator
by IP cameras,
as shown
in by the mobil
both
kit parts at
high
speed in the
feeding station
and to assist
feeding
assembly
performed
Fig. 3.40, in
in the
order
to improve
safety.by the human operators, as in the case
concentrate on the asse
of automatic operation in cooperation with operators in the
in comparison with con
assembly station. If the manipulators are to shift from parts feeding
This new cell produ
3.2.3 System
performance evaluation
and instead assist in the assembly being performed by the human
installed in manufactur
operators
according
to
the
progress
of
the
assembly,
higher
installation, the auth
To evaluate the effect of the new cell production system, a group of operatorsthe
was
productivity can be achieved.
cost of the setup includi
required to execute the assembly of a cable harness. In this task, operators were
To evaluate the safety management of the entire system,
management, and striv
required to
insert
into the corresponding
holes
in thelisted
connector.
includingcables
the above-mentioned
method, the
authors
up 168
risks including reasonably foreseeable misuses, and set up
Acknowledgments
countermeasures according to the ‘‘Three Step Method’’ (inherently safe design measure, safeguarding or complementary
This study is suppor
protective measure, and information for use), as defined in ISO
Technology Developm
12100 [9] as a risk assessment method. Since some risks have yet to
projects of element tech
3.2. A NEW CELL PRODUCTION ASSEMBLY SYSTEM WITH
HUMAN-ROBOT COOPERATION
109
Two parameters were measured in the experiments; assembly time and error
ratio. These two parameters were compared between conventional manual assembly cell production (Exp I) and the new cell production (Exp II). Five novice
and five expert operators performed three assembly trails for both the cells.
From Fig. 3.41, the authors can see that the overall performance is better
(shorter assembly time) in the new cell production (Exp II). Both the novices and
the experts take almost the same time to complete the assembly, from the first trial
in the case of the new cell production (shown by the dotted line). This means that
the assembly can be executed in a short time even by novice operators. Comparing this with the assembly time required for conventional cell production (Exp I),
the novice operators need only 50% of the time in the new cell production (Exp II).
M. Morioka, S. Sakakibara / CIRP Annals - Manufacturing Technology 59 (2010)
12
That is, the productivity
is doubled. Note that
the assembly time in the third trial
converges to a minimum in all cases. This implies that the assembly operation
doubled. Note that the ass
is relatively easy and that a human’s ability to learn is high. In other words,
this
a minimum
in all cases. Th
relatively
system may be beneficial to situations where the product being manufactured
is easy and that a
words, this system may
frequently changed.
product being manufactu
In term of assembly quality, a 10% to 20% assembly error ratio (insertionIner-term of assembly qu
ror) was observed in conventional cell production (Exp I), while in the new
cell error) was obser
(insertion
production (Exp II) errors were totally prevented thanks to the assistance I),
of while
the in the new cell
prevented
robot, especially the laser pointer guidance and the instruction on the assembly thanks to the as
pointer guidance and the
sequences.
According to the exp
According to the experimental results discussed above, the authors canauthors
con- can conclude th
system
clude that the new cell production assembly system can accelerate the speed
at can accelerate the
Fig. 6. Safety management in the case of cooperation between operator and
while
which an operator
can work, while simultaneously preventing assembly errors. simultaneously pre
manipulators.
4. Conclusions and futu
In this study, the aut
assembly system, in whic
provided to the human op
with the robot while provi
performed. The safety m
from the viewpoint that
human operators in autom
can achieve higher produ
for the assembly of a cable
can be doubled relative t
tion. Furthermore, the au
with the new cell product
Fig.
7.
Difference
in
assembly
time
for
experts
and
novices.
Figura 3.41: Difference in assembly time for experts and novices.
because assembly errors
This new safety management enables the same manipulators to
In addition, the new
both kit parts at high speed in the parts feeding station and to assist
feeding by the mobile
in the assembly performed by the human operators, as in the case
concentrate on the assem
3.2.4 Conclusions
and future work
of automatic operation in cooperation with operators in the
in comparison with conve
assembly
station.
If
the
manipulators
are
to
shift
from
parts
feeding
In this study, the authors developed a new cell production assembly system,This
in new cell product
and instead assist in the assembly being performed by the human
installed in manufacturing
which physical
and information supports are provided to the human operators,
operators according to the progress of the assembly, higher
the installation, the autho
productivity can be achieved.
cost of the setup including
To evaluate the safety management of the entire system,
management, and strive t
including the above-mentioned method, the authors listed up 168
risks including reasonably foreseeable misuses, and set up
Acknowledgments
countermeasures according to the ‘‘Three Step Method’’ (inher-
110
CAPITOLO 3. THE ARTICLE SELECTED
reinforcing the safe cooperation with the robot while providing instructions on
the operations to be performed. The safety management for the cooperation is
new from the viewpoint that industrial robots can cooperate with human operators in automatic operation without stopping, which can achieve higher productivity. The authors demonstrated that, for the assembly of a cable harness, productivity with the new cell can be doubled relative to the conventional manual
cell production. Furthermore, the authors proved that the assembly quality with
the new cell production method can also be greatly improved because assembly
errors can be almost entirely eliminated.
In addition, the new method of automatic parts kitting and feeding by the
mobile manipulators allows the operators to concentrate on the assembly operations and improve productivity in comparison with conventional manual parts
kitting and feeding.
This new cell production assembly system is expected to be installed in manufacturing factories in the near future. To promote the installation, the authors
will make greater efforts to reduce the cost of the setup including the mobile
manipulators and the safety management, and strive to improve safety even
further.
Capitolo 4
R ISULTATI ,
LAVORI FUTURI ,
CONCLUSIONI
C
ON QUESTO CAPITOLO concludiamo la tesi esponendo in primo luogo i risul-
tati ottenuti. Successivamente vengono proposti alcuni suggerimenti per i
lavori futuri, consapevoli che molto lavoro rimane da fare in merito ai sistemi di
produzione ed assemblaggio flessibile. Infine vengono tratte alcune conclusioni.
Risultati
Negli ultimi dieci anni la ricerca nel campo dei sistemi flessibili di produzione ed
assemblaggio si è concentrata sia sugli aspetti di progettazione sia sugli aspetti
di funzionamento, con l’obiettivo, nella maggior parte dei casi, di massimizzarne
la produttività (Tabella 4.1).
Un sistema di produzione flessibile è caratterizzato da diverse macchine operatrici e diverse parti da manipolare, quindi, al fine di massimizzare la produttività (che di solito corrisponde a una minimizzazione dei costi di produzione)
è necessario allocare risorse e parti in maniera ottimale sincronizzando le varie
operazioni.
In pratica, l’allocazione ottimale delle risorse e i problemi di sincronizzazione,
dipendono strettamente dalla specifica architettura. Questo comporta una diversa modellazione a seconda del layout considerato con la conseguente formalizzazione della funzione obiettivo.
Sono stati proposti una vasta gamma di modelli: a partire da quelli stocastici (Mehmet Savsar et al. 2008, Majid M. Aldaihani et al. 2005, F.F. Uusitalo
et al. 2004), alle procedure euristiche (D. Battini et al. 2006, Xinyu Shao et al.
2005, Alessandro Agnetis et al. 2004), a quelle iterative (R. Manzini et al. 2004),
ai modelli analitici (Mehmat Savsar 2006, J. Krüger et al. 2005), agli algoritmi
dinamici (Weijun Zhang et al. 2005), ai modelli bi-criterio (Hacan Gultekin et
al. 2010, G. Fantoni et al. 2009), al problema del commesso viaggiatore (Hacan
111
Figura 4.1: Flexible Cell.
X.F. Zha
H. Du
Y.E. Lim.
2001
Design of a manufacturing
cell
2002
Modeling, design, and
planning of flexible robotic
assembly systems
Mini assembly cell
2004
Flexible manufacturing cell
Human–machine
cooperation
Flexible manufacturing cell
2004
Alessandro Agnetis
Arianna Alfieri
Gaia Nicosia
Flexible manufacturing cell
2004
2005
J. Krüger
B. Nickolay
P. Heyer
R. Manzini
M. Gamberi
A. Regattieri
A.Persona
F.F. Uusitalo
H. Viinikainen
R. Heikkilä
Mauricio CabreraRios
Clark A. MountCampbell
Shahrukh A. Irani
2005
Majid M. Aldaihani
Mehmet Savsar
Flexible assembly cell
Flexible assembly cell
2005
2006
Mehmet Savsar
Manufacturing cell
Mixed-model assembly
2008
Mehmet Savsar
Majid Aldaihani
Manufacturing cell
Human–machine
cooperation
2006
2009
Xinyu Shao
Li Ma
Zailin Guan
D. Battini
M. Faccio
E. Ferrari
A. Persona
F. Sgarbossa
Weijun Zhang
Theodor Freiheit
Huashu Yang
2009
J. Krüger
T.K. Lien
A. Verl
In-line robotic cells
2009
In-line robotic cells
Robot centered cells
2010
Hakan Gultekin
M. SelimAkturk
Oya Ekin Karasan
Human–machine
cooperation
Object
2010
2010
M. Morioka
S. Sakakibara
Serdar Yildiz
Oya Ekin Karasan
M. Selim Akturk
Hakan Gultekin
Oya Ekin Karasan
M. Selim Akturk
Year
Authors
One machine and at most
k parts
Different machines work
together within a common
workspace
Two- machine
One-robot
Different machines work
together within a common
workspace
Multi-station rotating table
Two- machine
One-robot
Parallel non-identical CNC
machines
Hybrid assembly
Workplace and time
sharing systems
Interfaces between human and machines
Robotics
Sensors and actuators
Safety systems
Vision systems
Vision systems
Manual operators
Actuators
Identical parts
Identical CNC machines
M-machine
One-robot
n° of tasks and n° of
components
Different parts
Flexible gripping devices
Vibratory bowl feeder
Elevator feeder
Belt feeder
Drum feeder
Flexible component feeders
Flexible automated logistics
Intelligent warehousing
Identical parts
Identical CNC machines
M-machine
One-robot
Identical parts
Identical CNC machines
Parts feeding station
Assembly station
Safety management technology
Double manipulator robot
Multi-modal assembly table with horizontally
built-in LCD monitor
Laser pointer
Laser scanner technology or vision systems for
3D workspace surveillance
Hybrid assembly
Workplace and time
sharing systems
Two- machine
One-robot
Cell Features
Technologies
Composition of the cell
Petri nets
Function-behaviorstructure model
Simulation
Design of experiments
Regression analysis
Taguchi methods
Deterministic methods
Stochastic methods
Iterative procedure
Heuristic algorithm
Analytical models
Stochastic model
Markov chain
Petri nets
Heuristic procedure
Profit value
Robustness
Minimize setup costs
Minimize FMC cost per unit
Minimize the completion
time for a single product or
a batch of products
Load-balancing
Production output rate
Optimal pallet capacity
Robot speed
Exact
numerical solutions
Economic analysis
Design, simulation,
analysis and evaluation,
and even layout
optimization of the flexible
assembly system
Reduce the development
time and cost of assembly
systems
Several feasible and
potentially profitable
designs
Economical hardware
platform
Flexible safety area
Optimally allocate
resources to parts
Synchronize the various
operations in the system
Reduce both the
complexity of the assembly
process and assembly
costs
Schedule tasks
Dynamic programming
algorithm
Optimize the productivity
Optimize the average cycle
time
Corrective maintenance
Preventive maintenance
Opportunistic maintenance
Simulation
Analytical models
Resource allocation
Improve productivity
Improve occupation
Cost reductions per unit
Reduced investment
increase reuse value for
new product
Optimal number of
machines
Allocation of the operations
to the machines
Processing times of the
operations on the
machines
Robot move sequence
Improve productivity
Output
Analyze performance
Reliability analysis
Availability analysis
Distance matrix consists of
decision variables
Cycle time analysis
Exact solution procedure
Heuristic algorithm
Solution procedure
Stochastic model
Minimize the cycle time
Minimize the total
manufacturing cost
Minimize the cycle time
Minimize the total
manufacturing cost
Minimize the cycle time
Minimize assembly time
Minimize error ratio
Objective function
Optimize the productivity
Best machining rates
Robot loading and
unloading rates
Pallet capacity
Pallet transfer rates
Optimize machine repair
rates
Heuristic procedure
Multiagent framework
Bidding mechanism
Travelling salesman
problem
Bicriteria model
Bicriteria model
Model
112
CAPITOLO 4. RISULTATI, LAVORI FUTURI, CONCLUSIONI
113
Gultekin et al. 2009), alle reti di Petri (Weijun Zhang et al. 2005, X. F. Zha et al.
2001), al disegno degli esperimenti unito all’analisi di regressione e al metodo di
Taguci (Mauricio Cabrera-Rios et al. 2002) passando per la simulazione (Mehmat
Savsar 2006, Mauricio Cabrera-Rios et al. 2002) si sono ottenute soluzioni sia esatte che approssimate. Si deve comunque sottolineare che in tutti i casi analizzati
le soluzioni approssimate hanno descritto con un ottimo grado di precisione le
applicazioni pratiche consentendo un notevole risparmio computazionale.
Interessanti risultano gli studi che riguardano i sistemi ibridi caratterizzati
dalla cooperazione tra uomo e robot con l’obiettivo di migliorare oltre alla produttività anche la versatilità del sistema produttivo. Infatti il loro minore grado di
automazione rispetto ai sistemi completamente automatizzati, ne conferisce una
migliore adattabilità nel passaggio da una produzione all’altra. I sistemi ibridi rappresentano una forma di razionalizzazione, che attraverso l’integrazione
delle risorse umane nel ciclo produttivo possono contribuire positivamente anche
all’occupazione.
Importanti sono i risultati emersi dal punto di vista della convenienza economica. Nello studio condotto da J. Krüger et al. 2009 sono stati valutati i costi
unitari di produzione in funzione delle quantità, nelle tre configurazioni: automazione ibrida, cella robotizzata e linea di trasferimento automatizzato. In
base a quanto evidenziato, l’automazione ibrida grazie ai contenuti investimenti che la caratterizza, ne viene consigliato l’utilizzo in presenza anche di lotti di
produzione di piccole e medie dimensioni. Basandosi sull’esempio PowerMate
(Schrft RD et al 2005) sono stati valutati i costi dell’automazione ibrida attraverso
il calcolo del Valore Attuale Netto dei flussi di cassa nelle configurazioni pura
cella robotizzata, assemblaggio manuale ed assemblaggio ibrido, assegnando a
quest’ultimo la soluzione migliore. Il caso trattato si basa su varie ipotesi, ad
esempio, che tutte le alternative producono la stessa qualità e che tutte le attività
possano essere eseguite sia da un robot sia da un essere umano. Inoltre gli aspetti
attinenti alle malattie e alle vacanze degli operatori non sono stati presi in considerazione. Tuttavia, questo esempio mostra una forte riduzione del tempo di
lavorazione nel caso di un sistema ibrido rispetto alla soluzione manuale e alla
soluzione automatica. Pertanto sistemi ibridi possono risultare economicamente
vantaggiosi.
Lotter e Wiendahl (2006) hanno esaminato le prestazioni dei sistemi di montaggio ibridi, che comprendono l’alimentazione e l’approvvigionamento automatico dei pezzi. Evidenziando che il tempo di tali attività non direttamente
utilizzato per l’assemblaggio può essere notevolmente ridotto incrementando di
fatto l’efficienza complessiva del processo di produzione. Questo tipo di montaggio ibrido è stato confrontato con l’assemblaggio completamente automatizzato
per prodotti diversi. In sei dei sette casi esaminati i costi unitari di un impianto
ibrido risultarono inferiori a quelli di un impianto completamente automatico.
114
CAPITOLO 4. RISULTATI, LAVORI FUTURI, CONCLUSIONI
Sulla base dell’esperienza ottenuta in compiti di assemblaggio con dispositivi
di automazione intelligente (IAD), Stanley Automation stima una riduzione dei
costi degli infortuni, dei costi di gestione e costi del lavoro pari al 58% sul totale.
Nello studio proposto da M. Marioca et al. nel 2010, gli autori hanno scelto di realizzare una nuova cella di produzione costituita da tre sottosistemi; (A)
una stazione di alimentazione delle parti componenti, (B) una stazione di assemblaggio e (C) la tecnologia di gestione della sicurezza. Essi hanno dimostrato
che, per il montaggio di un fascio di cavi, la produttività con la nuova cella può
essere raddoppiata rispetto alla produzione convenzionale (cella manuale). Inoltre la qualità dell’assemblaggio con il nuovo metodo può essere notevolmente
migliorata, eliminando quasi del tutto i difetti.
Ulteriori applicazioni economiche delle soluzioni ibride si potrebbero verificare in presenza di situazioni in cui l’assemblaggio non può o può essere fatto
solo con alti sforzi manuali o con soluzioni complesse automatizzate.
Le tecnologie chiave che governano questi sistemi sono:
1. Le interfacce tra l’uomo e la macchina che garantiscono una efficace cooperazione tra le parti, si distinguono in:
• interfacce remote come le interfacce visive, interfacce per gesti e la
voce;
• interfacce fisiche come le interfacce aptiche, display e caschi con display (HMDs), nonché sistemi di force feedback.
2. La robotica con:
• I robot assistenti: dispositivi flessibili ad interazione diretta attraverso
sensori, attuatori e procedure;
• I robot di collaborazione o cobots: dispositivi meccanici su guida dotati
di servomotori, dove un operatore umano fornisce forza motrice;
• La robustezza e stabilità dell’interazione uomo-robot;
• L’apprendimento interattivo dell’assistente robot;
• I sistemi robotici umanoidi antropomorfi;
• I robot portatili.
3. I sensori e gli attuatori. Oltre a strategie di controllo avanzato per realizzare un robot dal comportamento conforme e quindi sicuro un secondo approccio favorisce l’impiego di attuatori per garantire la sicurezza durante
l’interazione uomo-robot;
4. La gestione della sicurezza. La coesistenza di sistemi robotici con gli esseri
umani nello stesso dominio fisico/temporale, pone il problema fondamentale di garantire la sicurezza attraverso:
• Sistemi di pre-collisione - limitazioni e controlli basati su sensori di
sorveglianza dell’area di lavoro:
115
– Definendo delle zone limitate in cui i bracci robotici si possono
muovere (poco flessibile);
– Utilizzando tecnologie di rilevamento scanner in grado di determinare quando un operatore entra nella zona di influenza del robot.
In funzione della posizione relativa, si procede o ad un arresto di
emergenza o ad un rallentamento dell’operatività;
• Sistemi di post-collisione - sensori integrati nel robot per individuare
rapidamente la collisione tra l’uomo e la macchina.
5. L’approvvigionamento dei componenti e la logistica. L’efficienza complessiva dei processi flessibili non dipende soltanto dal rapporto ottimale
tra l’umano e l’automazione, ma anche da come i componenti sono alimentati ai luoghi di lavoro cooperativo. Poiché la cooperazione uomo-macchina
ha come obiettivi una maggiore flessibilità e adattabilità, il processo di alimentazione delle parti deve essere altamente flessibile. Si è riscontrato che
nelle soluzioni proposte per l’automatizzazione del sistema produttivo circa il 75% del costo totale è da imputare agli alimentatori e ai sistemi di
trasporto che consentono di automatizzare la logistica della cella, mentre le
apparecchiature per la produzione e il montaggio incidono per il restante
25%. La vera sfida è quindi l’alimentazione flessibile delle parti componenti. Soluzioni di automazione per l’alimentazione e la logistica esistono da
lungo tempo, basta pensare alla catena di montaggio di Henry Ford per la
sua automobile modello T. Ma la maggior parte di queste soluzioni sono
rigide, valide per la produzione di massa. L’alimentazione automatica è
risultata essere piuttosto impegnativa a causa della grande varietà di pezzi.
Ancora oggi non esiste una teoria generale o metodologia per determinare
la soluzione all’automazione dei componenti. Dedicheremo a tal proposito
una discussione più dettagliata successivamente. La sfida della flessibilità
inizia già nella fase di presa. I robot sono intrinsecamente molto flessibili,
ma la loro performance è spesso limitata dalla capacità di afferrare l’oggetto
che deve essere gestito. Pinze in grado di gestire la varietà di consistenza
dei materiali da quelli duri a quelli molli e flosci non sono comunemente in
uso nell’ambito industriale e la tecnologia in questo settore è ancora in via
di sviluppo.
Il rifornimento degli alimentatori è un’altra sfida. Il funzionamento automatico dei sistemi di montaggio richiede un flusso costante di componenti.
Tradizionalmente questa sezione del sistema di produzione è stata automatizzata in misura minore, in quanto la frequenza delle operazioni è inferiore
e di conseguenza può sembrare meno conveniente intervenire in quest’area.
Ma anche in questa zona ci sono sviluppi tecnici che puntano verso una
maggiore flessibilità di soluzioni di trasporto automatico come i nuovi veicoli a guida automatica (AGV) che integrano sistemi di navigazione au-
116
CAPITOLO 4. RISULTATI, LAVORI FUTURI, CONCLUSIONI
tomatica svincolandoli da percorsi fissi, rendendoli di fatto più efficaci e
flessibili ed applicabili ad una gamma molto più ampia di sistemi di montaggio. I sistemi di trasporto intelligenti sono un’altra importante novità. I
trasportatori sono passati dai sistemi a nastro semplice con una velocità di
trasporto costante ai sistemi complessi di trasporto che portano gli oggetti
lungo percorsi che sono adeguati alle esigenze di un processo produttivo
specifico. Sistemi frequentemente utilizzati prevedono una sorta di identificazione del pallet su cui viaggiano i componenti così da determinarne
l’instradamento corretto lungo il percorso di trasporto. É quindi possibile
mescolare prodotti diversi nelle diverse fasi del processo di produzione su
un unico sistema di trasporto.
Un discorso a parte meritano i sistemi di alimentazione Tabella 4.2. In molte
applicazione industriali l’alimentazione delle parti è affidata ai vibrotrasportatori
che hanno il compito di separare i componenti stoccati alla rinfusa convogliandoli
secondo un orientamento prestabilito al punto di prelievo. I vibroalimentatori
sono caratterizzati da bassa velocità di processo e bassa affidabilità a causa degli
inceppamenti dei componenti da orientare. A causa di questi guasti tutta la linea
deve essere fermata generando dei costi elevati. Ogni vibroalimenatore è specifico per la parte da alimentare per cui è stato progettato, così come i suoi dispositivi di orientamento che devono essere adattati nel processo di installazione in
modo da ottenere l’output desiderato. L’elettromagnete è stato e viene comunemente usato come mezzo per generare la vibrazione in queste apparecchiature.
Tuttavia, a causa della complessità della sua struttura meccanica presenta diversi
problemi, come il rumore, il movimento non lineare delle parti e così via. Per risolvere questi inconvenienti S. B. Choi et al. 2004 hanno proposto di utilizzare un
piezzo attuatore per generare la vibrazione. I vantaggi del materiale piezoelettrico includono un rapido tempo di risposta, una larga frequenza di banda ed una
capacità di controllo accurato. Alcuni lavori di ricerca si sono concentrati sulla
modellazione della dinamica delle parti. Mentre J.A. Vilán Vilán et al. 2009 si
sono limitati unicamente al regime puramente periodico, I. Han et al. 2002 hanno
affrontato anche il regime caotico, valutando gli effetti che la variazione di alcuni
parametri fisici del vibroalimentatore hanno sul comportamento dinamico e sulla velocità di avanzamento delle parti. È stato verificato sperimentalmente che il
tasso di trasporto in regime caotico è più o meno indipendente dalle variazioni
dei parametri esterni. Z. Hu et al. nel 2005 hanno esaminato l’effetto del disaccoppiamento del moto lungo gli assi x ed y per l’alimentazione di parti fragili e
delicate. Il controllo separato della vibrazione permette di ottenere due vantaggi: il primo luogo si riesce a controllare meglio il moto delle parti, mantenendole
in contatto con la pista vibrante anche se la velocità di avanzamento è elevata,
questo si traduce in una buona protezione delle parti fragili; in secondo luogo
attraverso opportuni feedback si può controllare la forma d’onda che origina la
2001
2001
2000
1997
2011
N.F. Edmondson
A.H. Redford
Petar B. Petrović
Vladimir R. Milačić
Gvozden M. Dželatović
Figura 4.2: Techniques for flexible feeding.
Wendy Wolfson
Steven J. Gordon
G. Rosati
M. Faccio
A. Carli
A. Rossi
Proceeding:
Linear vibrating
Belt conveyor
Electromagnets
Different material
Convenience analysis and validation of a fully flexible assembly system
16th IEEE International Conference on Emerging Technologies and Factory Automation
Flexible parts feeders
Modular design
Belt feeder
Vibratory bowl feeder
Small parts
Small parts
Wide range of parts
Passive odules
Vision systems
Flexible manipulator
Passive and active modules
Computer algorithm to supply an
air jet pulse
Sensor
Sensor technology
Standard non-active orientation
blade
Method for handling cylindrical
parts
Pushing
Tilted plates
Curved tips
Three-dimensional generalization
of conveyor belts with fences
2002
Standard modules
Decoupled vibratory feeder
Vision system
Sensor
Air jet system
Air streams
Aerodynamic effects
Standard modules
Electro-pneumatic cylinders
Stepper motors
Vision systems
Anthromopic Robot
Compliance device
Small engineering parts
Wide range of parts
Variable
Electric field
Flexible parts feeders
Polyhedral parts
Different material
Mini and Microparts
Tiny parts
Physically robust
Complex geometries
Weight
Size
Local air resistance
Projected shape
Center of gravity
De-couple the X and Y components
vibratory
Conductive and dielectric
objects
Heavier load
De-couple the X and Y components
vibratory
Vision systems
Anthropomorphic robot
Vision systems
Robot
Complex geometries
Standard modules
Weight
Roughnes
Features
Dynamic
Geometric
Electromagnetic
Modules of the ring system
Array of microaztuator
Oriented by
Vibratory bowl feeder
Electromagnets
Electromagnets
Different material
Family of similar parts
Delicate parts
Fragile parts
Steel
Glass
Plastic
Microparts
Variable
Small parts
Screw
Nuts
IC components
Semi-conductor industry
Different material
Small parts
Microparts
Parts Size
Metallic subassemblies
Delicate
Fragile
Parts Type
2002
Vibratory feeder
Electromagnets
Piezoceramic actuator
Electromagnets
Piezoelectric actuator
Electrostatic fields
Vibrating platform
Bimorph piezoelectric actuators
Piezoelectric actuator
Belt conveyor
Parallel bimorph beams
Piezoelectric actuation
Electromagnetic
Electromagnets and leaf springs
Vibratory drive
2002
Nebojsa I. Jaksic
Gary P. Maul
I. Han
Y. Lee
N. F. Edmondson
A. H.. Redford
Robert-Paul Berretty
Mark H. Overmars
A. Frank van der
Stappen.
2003
Vibratory bowl feeder
2003
L. Han
S.K. Tso
Vibratory bowl feeder
2004
Modular design
Vibratory bowl feeder
2005
2003
Vibratory bowl feeder
Modular feeder
2005
2005
Meander-line structure
2005
H.-P. Wiendahl
A. Rybarczyk
Z. Hu
G.P. Maul
D. Farson
M.L. Tay
Patrick S.K. Chua
S.K. Sim
Y. Gao
S.B. Choi
D.H.Lee
M. H. Jiang
Patrick Chua
F. L. Tan
G. Fantoni
M. Santochi
Linear vibratory
Machine vision system
2006
2006
Z. Hu
D.F. Farson
G.P. Maul
Yung Ting
Ho-Chin Jar
Chung-Yi Lin
Jeng-Shen Huang
Andrew Perks
Machine vision system
2008
Piezoelectric part feeders
Horizontal platform vibrated
Vibratory bowl feeder
2009
2007
Vibrator BRUSH
Modular feeder (Vibratory BOWL
feeder and a ring system)
2010
Feeder
2010
Year
Paul C.-P. Chao
Chien-Yu Shen
G. Fantoni
M. Santochi
G. Reinhart
M. Loy
J.A. Vilán Vilán
A. Segade Robleda
P.J. García Nieto
C. Casquerio Placer
G.Rosati
G. Boschetti
A. Biondi
A. Rossi
Authors
Axiomatic Design Theory
Chaotic regime
3D simulation software
Modal analysis
Finite element method
Neural network (Backpropagation, ART2, ARTMAP)
Force analysis
One-dimensional hardware model
Control algorithm
Rayleigh–Ritz decomposition
method
Lagrange’s equations
Numerical
simulations
Newton Method
Spread Sheet
Structural mechanical model
Lagrangian Equations
ANOVA
Analysis Tools
117
118
CAPITOLO 4. RISULTATI, LAVORI FUTURI, CONCLUSIONI
vibrazione ottimizzando le performance del vibroalimentatore. M. L. Tay et al.
2005 hanno sviluppato un alimentatore convenzionale flessibile e programmabile. Con l’ausilio di un sistema di visione e di un computer, si controllano dei
cilindri elettro-pneumatici e dei motori passo passo per orientare correttamente
le parti componenti senza doverle respingere come avviene solitamente. Il riconoscimento dell’orientamento delle parti avviene tramite reti neurali. Questo
sistema estende l’applicabilità dei vibroalimentatori convenzionali a famiglie di
parti simili senza costi aggiuntivi di set-up. Interessante l’approccio di Nebojsa
I. Jaksic et al. 2001 dove si descrive lo sviluppo di un modello di comportamento di una parte componente da riorientare con un getto d’aria controllato da un
computer. L’algoritmo di controllo accetta il peso della parte, la geometria e il
suo attuale orientamento. Dei sensori confrontano l’orientamento presente con
quello voluto e l’algoritmo determina l’impulso adeguato del getto d’aria. Nel
2003 M. H. Jiang ha sviluppato un software di simulazione 3D basato su modello matematico, in grado di eseguire la simulazione dinamica del movimento di
una parte componente e dell’interazione della stessa con i meccanismi di orientamento. In questo modo si riesce a valutare la bontà del progetto dell’alimentatore prima di realizzarlo fisicamente riducendo gli extra costi derivanti da errata
progettazione.
Oltre agli studi sui vibroalimentatori, lavori di ricerca si sono dedicati allo
studio di alimentatori modulari. G. Reinhart et al. 2010 presentano un metodo di
progettazione che assicura un funzionamento ottimale attraverso l’adattamento
geometrico. Il metodo si basa su un modello strutturale del sistema meccanico,
che è stato derivato dalle equazioni di Lagrange. G. Fantoni ed M. Santochi nel
2005, descrivono lo sviluppo di un alimentatore per parti mini e micro basato su
campi elettrostatici. La movimentazione e l’orientamento avviene grazie ad una
piattaforma vibrante e ad un campo elettrico. La vibrazione viene utilizzata per
ridurre le forze di attrito tra il piano ed i componenti da orientare. In aggiunta, le
operazioni di set up risultano molto veloci dato che il sistema si compone di moduli standard facilmente configurabili. H.-P. Wiendahl et al. 2003 si pongono due
principali obiettivi nella realizzazione del loro alimentatore modulare: flessibilità
e velocità. Obiettivi che raggiungono grazie ad un approccio aerodinamico. Gli
autori sottolineano che le caratteristiche della parte componente che influenzano
il suo orientamento tramite il getto d’aria sono: la resistenza aerodinamica, la
proiezione della forma e il centro di gravità. Nell’alimentatore proposto si utilizza un approccio composto dai seguenti passi: (i) singolarizzazione dei pezzi, (ii)
preorientamento dei pezzi, (iii) orientamento finale con procedura aerodinamica,
(iv) sistema di movimentazione dei pezzi. Ancora con un approccio modulare
Petar B. Petrović et al. nel 2000, risolvono il problema per l’alimentazione ad alta
velocità nel montaggio di piccole parti. Il sistema prevede una zona di stoccaggio, un nastro trasportatore o un piano vibrante che portano le parti nella zona
119
seguente di orientamento composta da moduli attivi e passivi.
G. Fantoni ed M. Santochi hanno presentato nel 2010 un sistema di alimentazione basato su delle setole opportunamente inclinate. Esso consente il movimento controllato e il corretto allineamento dei diversi componenti per forma,
peso e rugosità superficiale. L’alimentatore analizzato sia dal punto di vista teorico che sperimentale ha dimostrato una buona capacità di orientamento, sia in
configurazione passiva (solo scivolo) sia in quella attiva (alimentatore vibrante).
Sulla base di quanto fatto nel 2005 Z. Hu et al. nel 2006 hanno applicato il disaccoppiamento della componente vibratoria ad un alimentatore lineare. È stato
sviluppato un semplice ma efficace algoritmo di controllo per la gestione della
vibrazione orizzontale e verticale in modo da ottenere la traiettoria voluta. Per
consentire una applicazione commerciale alla ricerca è stato implementato un
attuatore economico del tipo PZT.
Utilizzando due attuatori piezoelettrici alimentati da corrente alternata, Yung
Ting et al. 2005 presentano un meccanismo in grado di trasportare componenti. Il modello dinamico, il movimento, la traiettoria ottimale e la velocità di
avanzamento vengono studiati analiticamente e verificati sperimentalmente.
La progettazione meccatronica di un sistema di alimentazione flessibile presentata da L. Han ed S. K. Tso nel 2003 prevede: un vibroalimetatore disaccoppiato, un sistema di visone, dei sensori di spostamento, degli amplificatori di potenza il tutto supportato da un sistema informatico. I parametri di funzionamento
come l’angolo di vibrazione, l’accelerazione e la differenza di fase possono essere regolati dal software online. Questo permette all’alimentatore di gestire una
vasta gamma di parti senza una conversione fisica. Possono essere alimentate
anche parti di materiale diverso con rendimenti sempre vicini all’ottimo. Infine
il sistema di visione può individuare facilmente eventuali inceppamenti e quindi
ordinare al software di invertire il senso di avanzamento risolvendo il problema.
N. F. Edmonson e A. H. Redford nel 2001 hanno presentato un alimentatore a
nastro flessibile, in grado di alimentare geometrie complesse usando una tecnologia a sensore, moduli standard per l’orientamento passivo e un metodo per la
gestione delle parti cilindriche. Le caratteristiche del prototipo realizzato sono:
basso costo, design compatto, set-up veloce e semplice e capacità di alimentare
pezzi con velocità di avanzamento tra 30 e 60 pz/minuto.
Robert Paul Berretty et al. 2002, hanno dimostrato che è possibile orientare
parti componenti poliedriche asimmetriche attraverso una sequenza di spinte/urti senza l’utilizzo di sensori, indipendentemente dal loro orientamento iniziale
utilizzando una mascella composta da due piani ortogonali.
Per concludere riportiamo quelli che sembrano a nostro avviso i lavori più
interessanti nell’ambito dell’alimentazione flessibile delle parti componenti. La
flessibilità nel rispondere alle richieste del mercato è un requisito fondamentale
per i produttori che ricercano un vantaggio competitivo in un’economia globa-
120
CAPITOLO 4. RISULTATI, LAVORI FUTURI, CONCLUSIONI
le. Prodotti sempre più differenziati con cicli di vita brevi, bassi volumi e una
riduzione dei tempi di consegna al cliente, impongono un miglioramento delle
performance operative. L’introduzione di un nuovo prodotto o il cambiamento di un componente, comporta grandi modifiche nel processo di automazione.
Nella maggior parte dei casi si dovrà intervenire sulle apparecchiature con cui
i componenti vengono alimentati alle stazioni di assemblaggio, modificandone
nella migliore delle ipotesi il set-up arrivando nei casi più estremi alla completa
sostituzione delle stesse. Diverse sono le soluzioni proposte, (G.Rosati et al. 2008,
A. Perks 2006, F.F. Uusitalo et al. 2004, N. F. Edmondson et al. 2002, W. Wolfson
et al. 1997), tutte sostanzialmente basate sulla combinazione di un sistema di
visione ed un robot manipolatore. Il sistema di visione cattura l’immagine dei
componenti alimentati in modo casuale, un software di riconoscimento opportunamente tarato ne individua l’esatta posizione trasmettendo le coordinate al
robot. I componenti vengono di fatto alimentati direttamente alla cella di assemblaggio senza la necessità di prevedere sistemi di singolarizzazione e pick up.
Questa funzionalità è di solito indicata come la visione robotica (VGR). VGR è
una tecnologia in rapida crescita ed è un modo per ridurre i costi di gestione
mantenendo la produzione anche nei paesi con un costo del lavoro più elevato.
Lavori futuri
La ricerca futura dovrà concentrarsi su diversi aspetti nell’assemblaggio cooperativo ibrido come la robustezza dei componenti del sistema per la sicurezza, la
riduzione dei carichi ed altro al fine di estenderne il campo di applicabilità.
Una nuova sfida sarà quella di migliorare la cooperazione tra uomo e la macchina non solo per un singolo essere umano, ma anche per un gruppo di lavoratori
che devono svolgere un compito comune sostenuto da una o più macchine nello
stesso momento.
La stretta collaborazione tra uomo e macchina, in futuro, potrà anche cambiare il modo in cui sistemi automatizzati saranno programmati. Si può prevedere
che una vasta gamma di sistemi automatizzati cinematici come i robot useranno
interfacce per una efficace modalità di apprendimento sulla base di una interazione diretta con gli operatori.
L’approccio orientato all’automazione umana potrà in generale portare a una
modifica dei principi di design per i robot. Oggi la progettazione dei robot industriali si focalizza sulla precisione. Il controllo tramite l’applicazione di sensori esterni è necessaria per una interazione stabile e sicura. In soluzioni a lungo termine la progettazione di robot a sicurezza intrinseca terrà conto sia della
sicurezza che dell’esattezza del controllo.
Una ulteriore sfida nell’assemblaggio sarà quello di tenere traccia della quantità di pezzi in tutti i punti di stoccaggio. Questa sfida diventa molto importante
121
quando il processo è automatizzato. Situazioni di mancanza di parti, in qualsiasi
punto richiederebbe un arresto immediato delle operazioni con la necessità di un
intervento manuale per ricostituire le scorte e riavviare il processo. Gli odierni
sistemi ERP sono ben sviluppati per gestire la contabilità generale delle aree di
stoccaggio principale, cioè materie prime, stock intermedi e prodotti finiti. Ma
il livello delle scorte di una qualsiasi di queste aree non è aggiornato in tempo
reale. Pur essendo disponibile la tecnologia per monitorare in tempo reale lo stato di tutti i punti di stoccaggio, non è ancor oggi disponibile sul mercato una sua
implementazione commerciale.
Con riferimento alle celle costituite da macchine identiche che producono
pezzi uguali servite da un robot, future linee di ricerca potrebbero generalizzare
lo studio al caso di celle alimentate da diverse tipologie di pezzi servite da più
robot.
Nel campo degli alimentatori di parti sicuramente gli sforzi si dovrebbero
concentrare sullo sviluppo di un sistema flessibile e programmabile di alimentazione. Fortemente auspicato dal mondo industriale, questo nuovo sistema flessibile
permetterebbe una riduzione dei costi di set-up attualmente affrontati ad ogni
cambio di produzione. Appropriata la strada intrapresa con i sistemi VGR, i
quali sfruttando l’innovazione tecnologica nel campo dei dispositivi di presa e
del riconoscimento 3D, permettono di progettare un manipolatore in grado di
riconoscere, afferrare e quindi porgere al meccanismo di assemblaggio la parte
componente desiderata.
Conclusioni
Nella corsa degli ultimi anni, i sistemi di assemblaggio automatizzati sono stati
fortemente migliorati. Il processo di assemblaggio al giorno d’oggi viene eseguito
con un multiplo di flessibilità, velocità e sicurezza rispetto a quanto è stato alcuni anni fa. Ma, a differenza delle apparecchiature per il montaggio, i sistemi di
alimentazione dei componenti sono stati poco modificati. I metodi esistenti di alimentazione non soddisfano completamente le esigenze dei moderni dispositivi
di montaggio in termini di velocità, sicurezza e neutralità nei confronti delle varianti di produzione. Essi attirano una piccola quantità di investimenti causando
al contrario un gran numero di problemi. In molti casi, le unità di alimentazione
limitano la potenzialità massima di tutto il sistema, in quanto non raggiungono i
tassi di approvvigionamento richiesti dalle unità di assemblaggio.
Per quanto riguarda i sistemi ibridi, da quanto emerso si può presumere che
la prossima generazione di robot interagirà direttamente con gli operatori umani
per la manipolazione dei prodotti. Lo stretto legame tra uomo e macchina nei
sistemi di produzione cooperativo utilizzerà sempre di più i punti di forza di entrambe gli attori. In particolare i sistemi automatizzati saranno sempre più carat-
122
CAPITOLO 4. RISULTATI, LAVORI FUTURI, CONCLUSIONI
terizzati per la possibilità di lavorare senza interruzione, svolgendo compiti che
richiedono forza e precisione il tutto con alta produttività. D’altra parte l’essere
umano continuerà a fornire le incomparabili capacità senso motorie risultando
ideale per la gestione delle attività complesse e manifestando la capacità di un
rapido adattamento alle sequenze di un nuovo processo.
In generale i vantaggi di una stretta cooperazione tra i sistemi intelligenti di
assistenza e gli operatori umani costituiranno in futuro la base per un sistema di
assemblaggio e produzione più flessibile e sostenibile.
B IBLIOGRAFIA
[1]
A NDREW P ERKS . Advanced vision guided robotics provide “future-proof” flexible
automation. Assembly Automation 26/3, Pages 216-220, (2006).
[2]
G. B OOTHROYD. Assembly automation and product design. Dekker Inc., New
York, (1991).
[3]
G. FANTONI , M. S ANTOCHI . An analysis of cyclic scheduling problems in robot
centered cells. Computers & Operations Research, (2010).
[4]
M AURICIO C ABRERA -R IOS , C LARK A. M OUNT-C AMPBELL , S HAHRUKH
A. I RANI . An approach to the design of a manufacturing cell under economic
considerations. Int. J. Production Economics 78, Pages 223-237, (2002).
[5]
A LESSANDRO A GNETIS , A RIANNA A LFIERI , G AIA N ICOSIA . A heuristic
approach to batching and scheduling a single machine to minimize setup costs.
Computers & Industrial Engineering 46,Pages 793-802, (2004).
[6]
X INYU S HAO , L I M A , Z AILIN G UAN . A market approach to decentralized control of a manufacturing cell. Chaos, Solitons and Fractals 39, Pages 2303-2310,
(2009).
[7]
G. FANTONI , M. S ANTOCHI . A modular contactless feeder for microparts. CIRP
Annals - Manufacturing Technology, Volume 54, Issue 1, Pages 23-26, (2005).
[8]
M. M ORIOKA , S. S AKAKIBARA. A new cell production assembly system with
human-robot cooperation. CIRP Annals - Manufacturing Technology 59, Pages
9-12, (2010).
[9]
Y UNG T ING , H O -C HIN J AR , C HUNG -Y I L IN , J ENG -S HEN H UANG. A new
type of parts feeder driven by bimorph piezo actuator. Ultrasonics 43, Pages 566573, (2005).
[10] RH. H OLLIER. Automated materials handling. Ifs Splinger Verlag, (1983).
[11] M AJID M. A LDAIHANI , M EHMET S AVSAR. A stochastic model for the analysis of a two-machine flexible manufacturing cell. Computers & Industrial
Engineering 49, Pages 600-610, (2005).
123
124
BIBLIOGRAFIA
[12] J.A. V ILÁN V ILÁN , A. S EGADE R OBLEDA , P.J. G ARCÍA N IETO , C.
C ASQUEIRO P LACER. Approximation to the dynamics of transported parts in a
vibratory bowl feeder. Mechanism and Machine Theory 44, Pages 2217-2235,
(2009).
[13] H AKAN G ULTEKIN , M. S ELIM A KTURK , O YA E KIN K ARASAN. Bicriteria robotic operation allocation in a flexible manufacturing cell. Computers &
Operations Research, 37, Pages 779–789, (2010).
[14] I. H AN , Y. L EE. Chaotic dynamics of repeated impacts in vibratory bowl feeders.
Journal of Sound and Vibration, Volume 249, Issue 3, Pages 529-541, (2002).
[15] J. K RÜGER , T.K. L IEN , A. V ERL. Cooperation of human and machines in assembly lines. CIRP Annals - Manufacturing Technology 58, Pages 628-646,
(2009).
[16] W ENDY W OLFSON AND S TEVEN J. G ORDON. Designing a parts feeding system
for maximum flexibility. Assembly Automation, Volume 17, Number 2, Pages
116-121, (1997).
[17] D. B ATTINI , M. FACCIO , E. F ERRARI , A. P ERSONA , F. S GARBOSSA. Design
configuration for a mixed-model assembly system in case of low product demand.
Int J Adv Manuf Technol, Pages 188-200, (2006).
[18] G UNTHER R EINHART, M ICHAEL L OY. Design of the optimal feeding policy in an
assembly system. Journal of Manufacturing Science and Technology, Article in
Press, Corrected Proof, (2010).
[19] D. B ATTINI , M. FACCIO , A. P ERSONA , F. S GARBOSSA. Design of the optimal feeding policy in an assembly system. International Journal of Production
Economics, 121, Pages 233-254, (2009).
[20] G. FANTONI , M. S ANTOCHI. Development and testing of a brush feeder.
Manufacturing Technology 59, Pages 17-20, (2010).
[21] M.L. TAY, PATRICK S.K. C HUA , S.K. S IM , Y. G AO. Development of a flexible and programmable parts feeding system. International Journal of Production
Economics, 98, Pages 227-237, (2005).
[22] N EBOJSA I. J AKSIC , G ARY P. M AUL. Development of a model for part reorientation in vibratory bowl feeders with active air jet tooling. Robotics and Computer
Integrated Manufacturing 17, Pages 145-149, (2001).
[23] PAUL C.-P. C HAO , C HIEN -Y U S HEN . Dynamic modeling and experimental verification of a piezoelectric part feeder in a structure with parallel bimorph beams.
Ultrasonics, Volume 46, Issue 3, Pages 205-218, (2007).
BIBLIOGRAFIA
125
[24] W EIJUN Z HANG , T HEODOR F REIHEIT, H UASHU YANG . Dynamic scheduling in flexible assembly system based on timed Petri nets model. Robotics and
Computer-Integrated Manufacturing 21, Pages 550-558, (2005).
[25] Z. H U , D.F. FARSON , G.P. M AUL . Economic application of piezoelectric actuator
in linear vibratory feeding. Journal of Engineering Manufacture, Volume 220,
Number 12, Pages 1959-1967, (2006).
[26] M EHMET S AVSAR . Effects of maintenance policies on the productivity of flexible
manufacturing cells. Omega 34, Pages 274-282, (2006).
[27] G. R EINHART, J. W ERNER . Flexible Automation for the Assembly in Motion.
CIRP Annals - Manufacturing Technology, Volume 56, Issue 1, Pages 25-28,
(2007).
[28] N.F. E DMONDSON , A.H. R EDFORD . Flexible parts feeding for flexible assembly.
International Journal of Production Research, Volume 39, Number 11, Pages
2279-2294, (2001).
[29] R. MANZINI, M. GAMBER, A. REGATTIERI AND A. PERSONA.
Framework for designing a flexible cellular assembly system. International Journal
of Production Research, Volume 42, Number 17, Pages 3505-3528, (2004).
[30] N.F. E DMONDSON AND A.H. R EDFORD . Generic flexible assembly system
design. Assembly Automation, Volume 22, Number 2, Pages 139-152, (2002).
[31] X.F. Z HA , H. D U , Y.E. L IM . Knowledge intensive Petri net framework for concurrent intelligent design of automatic assembly systems. Robotics and Computer
Integrated Manufacturing 17, Pages 379-398, (2001).
[32] J. K RÜGER , B. N ICKOLAY, P. H EYER . Image based 3D Surveillance for flexible
Man-Robot-Cooperation. CIRP Annals - Manufacturing Technology, Volume
54, Issue 1, 2005, Pages 19-22, (2005).
[33] B. L OTTER L’automazione del montaggio. Tecniche nuove, Milano (1992).
[34] E. F ERRARI , A. PARESCHI , A. P ERSONA E A. R EGATTIERI. Logistica integrata
e flessibile - per i sistemi produttivi dell’industria e del terziaio. Esculapio, Bologna
(2009).
[35] L. H AN , S. K. T SO. Mechatronic design of a flexible vibratory feeding system.
Journal of Engineering Manufacture, Volume 217, Number 6, Pages 837-842,
(2003).
[36] J.J. U USITALO , H. V IINIKAINEN AND R. H EIKKILÄ. Mini assembly cell for the
assembly of mini-sized planetary gearheads. Assembly Automation, Volume 24,
Number 1, Pages 94-101, (2004).
126
BIBLIOGRAFIA
[37] S.B. C HOI , D.H. L EE. Modal analysis and control of a bowl parts feeder activated
by piezoceramic actuators. Journal of Sound and Vibration, 275, Pages 452-458,
(2004).
[38] M EHMET S AVSAR , M AJID A LDAIHANI. Modeling of machine failures in a
flexible manufacturing cell with two machines served by a robot. Reliability
Engineering and System Safety 93, Pages 1551-1562, (2008).
[39] P. B. P ETROVI Ć , V. R. M ILA ČI Ć , G. M. D IŽELATOVI Ć. New Feeding System
for High Speed Assembly of Small Parts. Manufacturing Technology, Volume
49, Issue 1, Pages 9-12, (2000).
[40] A. C ERINA. Organizzazione della produzione: guida ai moderni sistemi ed alle
nuove tecniche di analisi dei metodi produttivi. Buffetti, Roma (1988).
[41] H AKAN G ULTEKIN , O YA E KIN K ARASAN , M. S ELIM A KTURK. Pure cycles
in flexible robotic cells. Computers & Operations Research, 36, Pages 329-343,
(2009).
[42] H.P. TANG , T.N. W ONG . Reactive multi-agent system for assembly cell control.
Robotics and Computer-Integrated Manufacturing 21, Pages 87-98, (2005).
[43] R OBERT-PAUL B ERRETTY, M ARK H. O VERMARS , A. F RANK VAN DER S TAP PEN . Orienting polyhedral parts by pushing. Computational Geometry 21,
Pages 21-38, (2002).
[44] G. R OSATI , G. B OSCHETTI , A. B IONDI , A. R OSSI . On-line dimensional
measurement of small components on the eyeglasses assembly line. Optics and
Lasersin Engineering 47, Pages 320-328, (2009).
[45] Z. HU, G. P. MAUL AND D. FARSON. Piezo actuated vibratory feeding with
vibration control. International Journal of Production Research, Vol. 45, No. 5,
Pages 1089-1100, (2005).
[46] M. H. J IANG , PATRICK C HUA , F. L. TAN . Simulation software for parts feeding in a vibratory bowl feeder. International Journal of Production Research,
Volume 41, Number 9, Pages 2037-2055, (2003).
[47] R. A NTHONY, D. H AWKINS , DM. M ACRÌ , K. M ERCHANT. Sistemi di
controllo. McGraw-Hill (2001).
[48] H. R AMPERSAD . The DFA house, Assembly Automation, vol.13, n.4. (1993).
[49] H.-P. W IENDAHL , A. RYBARCZYK . Using air streams for part feeding systemsinnovative reliable solutions for orientation and transport. Journal of Materials
Processing Technology , 138, Pages 189-195, (2003).
BIBLIOGRAFIA
[50] C HRIS G REEN , J EFF K REUL . Vibratory Feeder Bowl Analysis, (2010).
127
R INGRAZIAMENTI
Sono passati 2 anni da quando mi sono seduto tra i banchi di Gestionale, anche se il
primo incontro con ingegneria risale al 1988. Sarà forse per questo che in pochi hanno
scommesso sulla mia Laurea? Così presto almeno?
Scherzi a parte, 23 sono i corsi frequentati per un totale di circa 2000 ore di lezione ed
almeno il doppio di studio. Le prove superate sono state una quarantina, tra scritti, orali,
elaborati, ecc. Da soli questi numeri basterebbero a sottolineare l’impegno richiesto, ma
se non siete ancora convinti credetemi quando dico che arrivare sin qui non è stato facile.
Analogamente non sarà facile ricordare tutte le persone da ringraziare e quindi mi scuso
sin d’ora se tralascio qualcuno.
Per iniziare un sincero grazie va al Professor Maurizio Faccio per gli insegnamenti
e la disponibilità dimostrata, ad Erika che non mi ha mai fatto pesare gli innumerevoli
week-end che abbiamo trascorso a casa perché dovevo studiare, in fondo questa laurea è
anche po’ sua!
Ovviamente non posso non salutare tutti gli amici dei corsi in particolare Fabrizio,
Erica ed Andrea che mi hanno aiutato in questi anni.
Infine ma non per questo meno importanti ringrazio i colleghi di lavoro Giovanna,
Susie di cui conservo tutti gli SMS inviati prima di ogni appello, Endri e Michele che
con la loro disponibilità e collaborazione mi hanno permesso di frequentare e sostenere gli
esami.
Per tutta la durata del corso ho tenuto il conto dei giorni che mancavano alla prova
successiva e al numero di prove che mancavano al traguardo finale. Ora è rimasta solo
la conta dei giorni che mi separano dalla seduta di Laurea, poi non ci sarà più nulla da
contare e sono certo che tutto questo un po’ mi mancherà.
129