Using Industrial Robots to Manipulate the Measured Object in CMM

ARTICLE
International Journal of Advanced Robotic Systems
Using Industrial Robots to Manipulate
the Measured Object in CMM
Regular Paper
Samir Lemes1,*, Damir Strbac1 and Malik Cabaravdic1
1 Mechanical Engineering Faculty, University of Zenica, Zenica, Bosnia and Herzegovina
* Corresponding author E-mail: [email protected]
Received 10 Oct 2012; Accepted 24 Apr 2013
DOI: 10.5772/56585
© 2013 Lemes et al.; licensee InTech. This is an open access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited.
Abstract Coordinate measuring machines (CMMs) are
widely used to check dimensions of manufactured parts,
especially in the automotive industry. The major
obstacles in automation of these measurements are
fixturing and clamping assemblies, which are required in
order to position the measured object within the CMM.
This paper describes how an industrial robot can be used
to manipulate the measured object within the CMM work
space, in order to enable automation of complex
geometry measurement.
Keywords Industrial Robot, CMM, Measurement
1. Introduction
In today’s global market, the importance of correct,
reliable and comparable measurements is the key factor
for achieving quality in activities and procedures in every
area of industry. Calibration, testing and measurement
are necessary elements in the development process or
progress in many disciplines of science and industry.
The majority of modern industrial measurements can be
categorized as GDT (Geometrical Dimensioning and
Tolerancing). Increasingly, measurements obtained by
coordinate measuring machines (CMM) are being used.
www.intechopen.com
CMMs are measuring devices with high measuring
speed. Positioning and rotation of the measured object
are always performed manually in the work area of the
coordinate measuring machine. The object, with defined
dimensions, shape and measuring surfaces, is measured
and controlled from several different sides. The measured
object needs to be positioned in certain positions relative
to the measuring device, which requires complex and
time-consuming actions. Each change of position also
requires a certain time, which can cause increased costs in
control and production processes. In order to reduce
these costs, several options for positioning a measured
object inside the working area of a CMM using an
industrial robot have been considered.
In a modern industrial environment the majority of
robots are automated systems controlled by computers.
Industrial robots have one or more robotic arms, control
devices with memory, and sometimes use sensors for
data acquisition. They usually support the manufacturing
process by positioning objects during machining or
welding, transportation, various technological operations,
automatic assembly, etc. They are also sometimes used
for pre-process, process and post-process control.
Industrial robots are widely used in processes which
require high quality and productivity.
Int.Strbac
j. adv.and
robot.
syst.,
2013, Vol.Using
10, 281:2013
Samir Lemes, Damir
Malik
Cabaravdic:
Industrial
Robots to Manipulate the Measured Object in CMM
1
Research along these lines has been conducted by
Santolaria and Aguilar [1]. They conducted a survey
about the development of kinematic modelling of robotic
manipulators and articulated arm coordinate measuring
machines (AACMM), taking into consideration the
influences of the chosen model on procedure parameters.
Their optimization algorithm included the terms linked to
the accuracy and repeatability of the procedure
presented. The algorithm follows the simple optimization
scheme of data obtained by investigation in several
spheres of objects placed at various positions within the
working area of both systems.
The majority of research mostly concentrates on
differences in the influences on measurement uncertainty
in Coordinate Measuring Machines. Lawford [2]
observed dysfunctional CMS software with unknown
measurement uncertainty and compared its influence
with measurement results. He also investigated the
prescribed testing algorithms and the checking of
industrial software, testing by comparison with given
algorithms. This yielded solutions to similar problems.
Fang, Sung and Lui [3] observed the influence of
measurement uncertainty from CMM calibration and
temperature in the working environment. If the
measurement uncertainty in CMM calibration is reduced,
this will also reduce the measurement uncertainty of the
machine itself. They also pointed out the importance of
temperature balance in the working environment before
the measurement is performed, i.e., the temperature
should be controlled in order to fit CMM working
specifications. CMM uncertainty can be reduced using
highly precise instruments such as a laser interferometer.
One of the first papers about CMM error compensation
was presented by Zhang et al. [4]. They described the
error compensation on bridge-type CMM, which resulted
in an improvement of accuracy by a factor of 10. They
also presented the correction of vector of equally
distributed points in the measured volume.
Software error compensation has been reported by a
number of authors. Ferreira and Liu [5], for example,
developed the analytical model for geometric errors of
the machining assembly; Duffie and Yang [6], meanwhile,
invented a method to generate the kinematic error
function from volumetric measurement error using a
vectorial approach.
Robot-CMM integration was performed for the first time
by the Mitutoyo Company [7]. They developed a software
module in order to adjust the actions of CMM and robotic
handling machines used for manipulation of measured
parts. Mitutoyo has released the source code in the hope
that third party software vendors will be able to use it as
a basis to develop products. To the authors’ knowledge,
2
Int. j. adv. robot. syst., 2013, Vol. 10, 281:2013
at the time of writing no achievements have been made in
this direction.
Hansen et al. [8] estimated measurement uncertainty of a
hybrid system consisting of an Atomic force microscope
attached to a coordinate measuring machine, using linear
combination of these two components. Although Hansen
et al. also combined two devices, their approach
nevertheless differs from ours: we use one system to
position the measured object, while they used a twocomponent system to perform the measurement.
Aggogeri et al. [9] used simulation and planned
experimentation to assess the measurement uncertainty
of CMMs. They identified and analysed five influence
factors, and showed that simulation can successfully be
used to estimate CMM uncertainty.
Weckenmann et al. [10] investigated how measurement
strategy affects the uncertainty of CMM results. They
defined the measuring strategy in relation to “operator
influence”, which has been neglected in other research.
This work showed that measuring strategy influences
CMM uncertainty, and that scanning capabilities of
modern CMMs, using significantly a larger number of
touch points, overcome this influence.
Wilhelm et al. [11] also investigated the influence of
measurement strategy, which they defined as the “task
specific uncertainty”. They also showed that virtual
CMM, using Monte Carlo simulation, can be used to
estimate uncertainty. Nevertheless, although these
authors mentioned that part fixture influences
uncertainty, they did not analyse this thoroughly.
Feng et al. [12] applied the factorial design of experiments
(DOE) to examine measurement uncertainty. They also
studied the effect of five factors and their interaction, and
showed that there is statistically significant interaction
between speed and probe ratio. They also showed that
uncertainty is minimized when speed is highest, stylus
length is shortest, probe ratio is largest, and the number
of pitch points is largest.
Piratelli-Filho and Giacomo [13] proposed an approach
based on a performance test using a ball bar gauge and a
factorial design technique to estimate CMM uncertainty.
They investigated the effect of length, position, and
orientation in work volume on CMM measurement
errors. The analysis of variance results showed a strong
interaction between the orientation and measured length.
Unlike the mentioned studies, the goal of the research
presented here was to assess whether robots can be used
to position the measured object in complex measuring
systems, using measurement uncertainty analysis and
estimating the factors affecting it.
www.intechopen.com
2. Problem deescription
Fig. 1 show
ws a typical assembly used to clamp
p the
uires
measured ob
bject in a CMM. Such a system requ
multiple meeasuring and
d clamping operations w
when
different feaatures of thee measured object are b
being
measured, eespecially if the
t
geometry
y is complex and
some portion
ns are inacceessible by CM
MM probes. SSuch
products sho
ould be measu
ured in multiplle steps.
Figure 1. Typiccal fixturing asssembly for CMM
M measurementts
The major time-consu
uming task in coordiinate
measuremen
nt of complex
x geometry with
w
CMMs iis to
choose the position of the
t
measured
d object and
d the
optimal comb
bination of prrobe stylii, wh
hich often preevent
the touch p
probes from coming
c
into contact with
h the
measured ob
bject. The obsscured surface
es are difficu
ult to
reach withou
ut repositionin
ng the measu
ured object. E
Every
repositioning
g of the meaasured objectt introduces new
error sourcess in the measu
urement. Our idea is to usee the
industrial rob
bot to manipu
ulate the measured object, i. e., to
position it au
utomatically.
were
easurements w
In order to ttest this possiibility, the me
conducted w
with three diffeerent systems:
1.
2.
3.
Measurin
ng fixed objecct with CMM only;
o
Measurin
ng with comp
plex-CMM-robot system , with
fixed maass of measureed object;
Measurin
ng with com
mplex CMM-ro
obot system, with
variable mass of meassured object.
on to use thee measurement uncertaintty to
It is commo
quantify thee quality off the measu
urement proocess.
Therefore, w
we decided to
t estimate itt for these tthree
systems, and
d to compare th
hem with the uncertainty oof the
robot and thee CMM as statted by the man
nufacturers.
2.1 Hypothesees
In order to p
prove whetheer it is possib
ble to use a roobot,
with structurrally and tech
hnologically liimited option
ns, in
procedures o
of precise meaasurement wiith a CMM, aat the
beginning off the research the following
g three hypoth
heses
were set:
www.intechopen.com
•
•
•
Hypothesis 1:
1 It is possiblee to use the first generation
n
robot with five degrees oof freedom fo
or positioning
g
measured object on CMM..
Hypothesis 2: Uncertain
nty of meassurement forr
complex CM
MM-robot meeasuring syste
em is within
n
limits of alllowed uncerttainty of mea
asurement off
CMM.
nty of meassurement forr
Hypothesis 3: Uncertain
complex CM
MM-robot meaasuring system
m depends on
n
measured object mass.
nalysing thee
These hypothesses were teested by an
asurement un
ncertainty of tthe complex CMM-robott
mea
mea
asuring syste
em, comparin
ng them wiith limits off
allowed measurin
ng uncertaintyy of CMM (hy
ypotheses 1 &
2) an
nd by varying
g mass of meaasured object to
t analyse thee
mea
asurement un
ncertainty wh
hen the masss is changed
d
(hyp
pothesis 3). Other
O
influen
nces, such ass geometricall
erro
or, deformatio
on, thermal eerror, measurring strategy,,
prob
be movement speed duringg measuremen
nt, measuring
g
dyn
namics, workp
piece propertiees, vibrations, temperaturee
chan
nge, etc., were
e not considereed in this rese
earch.
2.2 Objective
O
The primary obje
ective of this research was to open new
w
posssibilities in this field and too encourage im
mprovementss
in the capabilities of CMM
M machines in terms off
shorrtening the prrocedure and reducing the measurementt
cyclle duration. We
W tried to p
point out the possibility off
com
mbining differe
ent structurall solutions on
n modern and
d
preccise equipmen
nt in order too achieve fastt, reliable and
d
preccise measurem
ment, and thu
hus improve technical
t
and
d
tech
hnological cap
pabilities in industry, pro
oduction and
d
reseearch areas wh
here CMMs arre commonly used.
u
3. Ex
xperiment desscription
3.1 Equipment
E
used
d
The coordinate measuring
m
macchine Zeiss Co
ontura G2 7000
Aktiv with tactiile probing ssystem was used in thiss
urement ran
nge: 700x100
00x600 mm,,
reseearch (measu
mea
asurement un
ncertainty acccording to ISO 10360-2::
MPE
E_E = (1.8+L/3
300 μm, MPE__P = 1.8 μm).
a rotating oof the measured object wass
The positioning and
ually, inside the CMM’ss workspace,,
perfformed manu
whiich tended to take
t
a consideerably long tim
me. In order to
o
shorrten this time, to reduce th
he positioning
g error and to
o
min
nimize other errors, an ed
ducational rob
bot with fivee
degrees of fre
eedom was used: Rob
bot RV-2AJ,,
nufactured by Mitsubishi Ellectric–Melfa robots, Japan..
man
The measuremen
nt uncertainty of this robott is not stated
d
t manufactu
urer; the onlyy comparable parameter iss
by the
repeeatability, sta
ated to be ±00.04 mm. Th
he robot is a
statiionary roboticc system, with
h programmed
d motion path
h
and
d automatic determination oof the target.
Samir Lemes, Damir Strbac and Malik Cabaravdic: Using Industrial
Robots to Manipulate the Measured Object in CMM
3
3.2 Conditionss
3.4 Positioning
P
All measurem
p
wiith the condittions
ments were performed
and capacitties availablee at the la
aboratory at the
University o
of Zenica. The
T
tempera
ature during the
experiment was 21°C. The workp
piece and C
CMM
measuring ellements were cleaned prior to measurem
ment
in order to reemove possiblle contaminan
nts. There werre no
were
other machin
nes in the viicinity of the CMM; nor w
there any oth
her vibration sources
s
(excep
pt the CMM’ss and
robot's own
n vibrations). Prior to measurement,
m
the
calibration off the measurin
ng tools and measuring
m
sysstem
was perform
med using 25 mm ceramic reference sph
heres
manufactured by Zeiss, using
u
the calib
bration proceedure
defined by C
CMM softwaree Calypso.
The robot’s posittion compared
d to CMM was
w limited by
y
the robot’s arm-re
each limit or it
its workspace.. Accordingly,,
the robot was possitioned and fiixed in an opttimal position..
n
Thiss position wass defined by rrelative angullar rotation in
the arm’s joint, and
a
the positi
tion and orien
ntation of thee
ppers in the sp
pace, ensuringg correct performance of thee
grip
giveen assignmentt. The robot w
was attached to
t the CMM’ss
gran
nite table usin
ng Z-shaped p
profile elements with firm
m
screew connections.
ntrol of the robot
r
was seemiautomatic. The controll
Con
prog
gram (direct programming
p
g) for piece po
ositioning wass
follo
owed with manual
m
launch
hing, starting the program
m
for each
e
single measuring
m
phaase. After the robot
r
trapped
d
the measured object with the pneumatic grripper, it wass
n moved into a position enaabling measurrement with a
then
sing
gle stylus sysstem, makingg all geometrical featuress
easiily accessible by all touch
h probes assembled in thee
stylu
us system.
3.5 Geometrical
G
feaatures
Figure 2. Calib
bration with ceramic reference sphere.
s
3.3 Measured object
The measureed object wass selected to have geomettrical
features typ
pically found in coordinatte measurem
ments:
planes, conees, and cyliinders. The material of the
measured ob
bject was no
ot of great importance,
i
ssince
temperature deviations were neglig
gible (laboraatory
m
fo
orce (200 mN)) did
conditions), aand typical measurement
not deform tthe object. Thee object was made
m
of PVC
C and
the surface w
was metalized,, reducing surrface roughneess to
a minimum. Fig. 3 shows the
t four featurres measured.
Ø 42,5
Ø 39,0
a
b
c
7
d
12
R1
5
Ø 43,0
4
Ø 60,0
Ø 72,0
Dim
mensional me
easurements were repeate
ed a certain
n
num
mber of time
es on previiously described surfacess
defiining differen
nt workpiecce geometries. For each
h
partticular surface, the dimen
nsions were measured 255
timees, under the same conditioons, in order to compensatee
rand
dom errors. The number off measuremen
nts (the size off
the sample) was determined
d
acccording to th
he significancee
the probability
y of failing to
o
leveel of the test α = 0.01 and th
deteect a shift of one standard d
deviation β = 0.01
0 for a two-sideed test, assum
ming normall distribution and known
n
stan
ndard deviatio
on [14].
The planar featurres were meassured by sets of 250 pointss
distributed circularly, and conical features weree
asured by me
easuring two circles, each consisting off
mea
250 points, at distances of 1 m
m
mm (cone "b") and 3 mm
(con
ne "c") from th
he edges, in orrder to avoid filleted
f
edges..
The measuremen
nt results weere used to estimate thee
asurement unccertainty, as a measure of validity
v
of thee
mea
resu
ults and confirrmation of hyp
potheses 1 and
d 2.
ure was identical, but with
h
For hypothesis 3, the procedu
o measured oobject. The firrst measuring
g
incrreased mass of
cyclle was perforrmed on a C
CMM with th
he measuring
g
objeect fixed on th
he CMM’s graanite table. Th
he second and
d
third
d cycles were
e measured byy the complex
x CMM-robott
systtem.
ow the planess and cones used
u
to definee
Figss. 4 and 5 sho
the dimensions to
o be measured
d.
Figure 3. Meassured object witth defined geom
metrical featuress
(a - top plane o
of cone, b - coniccal portion with
h larger angle, c conical portion
n with smaller angle
a
and d - top
p plane of cylind
der)
4
Int. j. adv. robot. syst., 2013, Vol. 10, 281:2013
www.intechopen.com
Cone 1
Plane "a"
Plane "d"
Cone 2
Figure 4. Geom
metric features measured
m
by CM
MM.
Figu
ure 6. Robot arm
m position 1.
Figure 5. Meassured features in
n CMM softwarre.
The measureed dimensionss were defined
d as follows:
–
–
–
–
Diameteer d1 is the inteersection of plane ”a” and cone
1
Diameteer d2 is the inteersection of co
one 1 and conee 2
Diameteer d3 is the inteersection of plane ”d” and cone
2
Height H is the distan
nce between pllanes ”a” and ”d”.
Figu
ure 7. Robot arm
m position 2.
Sincce the measure
ed object’s maass was consta
ant during thee
firstt two measurring cycles, th
he mass of the
t
measured
d
objeect was increa
ased by addin
ng mass m = 600
6 g (Fig. 8)..
The third measurring cycle wass conducted with
w increased
d
ults were comp
pared with th
he first cycle.
masss and the resu
4. Experimen
nt
In the first measuring cy
ycle, the mea
asured object was
positioned an
nd fixed to th
he CMM’s measuring table,, and
in the second
d cycle the position
p
of the
e measured oobject
was defined by the robott’s arm positiion (i.e., auxiiliary
ding the meassured
elements in tthe robot’s arrm were hold
object) insid
de the CMM
M’s coordinate
e space. Betw
ween
me a
every single measurementt in this measu
uring cycle cam
m
fro
om one positioon to
phase of the robot’s arm movement
back again. Coordinates of the robot’s arrm in
another and b
both positio
ons were defined by the
e robot’s offf-line
programming in such a way that the robot could
d be
manually and that position memorized. A
After
positioned m
ot’s operating
this, the robo
g speed was defined.
d
Sincee the
robot repeateed this operattion for each measurementt, the
robot’s repeaatability was ±0.04 mm. Figs. 6 and 7 sshow
the first and tthe second rob
bot arm positiions, respectiv
vely.
www.intechopen.com
Figu
ure 8. Measuring
g object with ad
dditional mass.
5. Measurement
M
results
r
Mea
asurement results are shown
n in Table 1.
Samir Lemes, Damir Strbac and Malik Cabaravdic: Using Industrial
Robots to Manipulate the Measured Object in CMM
5
Measure-men
nt Measured vaalues (mm)
No.
Diameter Diameter Diameter
d3
d1
d2
1.
39.0485
42.5430
42.9402
4
2.
39.0482
42.5432
42.9400
4
3.
39.0484
42.5430
42.9401
4
4.
39.0483
42.5431
42.9401
4
5.
39.0485
42.5433
42.9404
4
6.
39.0484
42.5432
42.9405
4
7.
39.0484
42.5434
42.9407
4
8.
39.0486
42.5436
42.9408
4
9.
39.0486
42.5436
42.9410
4
...
...
...
...
23.
39.0481
42.5439
42.9405
4
24.
39.0481
42.5440
42.9404
4
25.
39.0487
42.5437
42.9409
4
Mean
valu
ue
39.0484
x1m
42.5436
42.9407
4
Standard
0.00026
0.00038
0.00043
0
deviation
Max
39.0487
42.5444
42.9415
4
Min
39.0478
42.5430
42.9400
4
Absolute
range E1
0.0010
0.0014
0.0015
Heigh
ht
H
18.80070
18.80065
18.80067
18.80068
18.80072
18.80070
18.80071
18.80072
18.80073
...
18.80071
18.80081
18.80074
18.80074
5.2 Statistical
S
analy
ysis
0.000050
18.80082
18.80065
The first step in statistical anaalysis was to question thee
mality of disstribution of the measurement results..
norm
Kurrtosis of all ressults was betw
ween -1.40 and
d 0.54, and thee
skew
w ranged betw
ween -0.99 annd 0.72. For 25
5 samples, thee
stan
ndard error off the skew is 0.49 and stan
ndard error off
the kurtosis is 0.9
98; therefore, bboth skew and
d kurtosis aree
d error, and we
w can assumee
lower than twice the standard
mal distributio
on of measureed data.
norm
0.00017
Table 1. Resullts of first measu
uring cycle - me
easuring object ffixed
on CMM's meaasurement tablee
5.1 Measurem
ment uncertaintyy
The declared
d measuremen
nt uncertainty of the CMM u
used
in this experriment is 1.800 μm. In all th
hree cases theere is
Type A stand
dard measurement uncertaiinty, which eq
quals
standard dev
viation times coverage
c
facto
or 2. The stan
ndard
measuremen
nt uncertaintiees of three me
easurement cy
ycles
are shown in
n Table 2 and Fig.
F 9.
Measured
value
Diameter d1
Diameter d2
Diameter d3
Height H
6
e that all meaasurements in the first casee
We can conclude
n the CMM’ss
exprressed lower uncertainty tthan stated on
calib
bration certifiicate. The meeasurement results
r
in thee
seco
ond case show
w that measu
urement uncerrtainty of thee
systtem CMM-ro
obot is signiificantly larg
ger than thee
decllared uncerta
ainty of the C
CMM. In the
e third cycle,,
wheere the CMM--robot system
m was used to
o measure thee
objeect with increased m
mass, the measurementt
uncertainties are larger than th
the declared uncertainty
u
off
n those in the
e second case..
the CMM, but still lower than
Thiss means that increased maass of the me
easured objectt
sligh
htly reduced uncertainty. T
This phenome
enon could bee
expllained by the increased inerrtia of the measured object,,
whiich stabilizes the system an
nd leads to more
m
accuratee
resu
ults.
o distribution
ns for measured values off
The histograms of
diam
meter d1 (Figss. 10 and 11) illustrate the normality off
distribution. The distributionss of most oth
her measured
d
ues have a sim
milar shape.
valu
Standard
d measurement uncertainty
u
(μm
m)
Case 1
Case 2
Case 3
only
CMMCMM-robot
C
witth
CMM
robot
added mass
0.53
12.33
3.86
0.76
12.98
4.69
0.86
10.92
3.95
1.00
9.72
4.17
Table 2. Stand
dard measuremeent uncertainty in
i three measurring
C
2. measurring on CMM-roobot
cycles: 1. meassuring only on CMM;
system; 3. meaasuring on CMM
M-robot, with ad
dditional mass.
Figu
ure 10. Histogram of distributioon of diameter d1
(Casse 2: CMM-robot) with fitting nnormal distributtion..
Figure 9. Comp
parison of meassurement uncertainties (μm).
Figu
ure 11. Histogram of distributioon of diameter d1
(Casse 3: CMM-robot + added mass)).
Int. j. adv. robot. syst., 2013, Vol. 10, 281:2013
www.intechopen.com
mea
asurement for complex CMM
MM-robot meassuring system
m
is within the limits of allowed un
ncertainty off
mea
asurement of CMM”)
C
shoulld be rejected, since P-valuee
for both one-tail and two-tail are significan
ntly lower thee
n critical value
e of t-variable for sample size 25, α being
g
than
eith
her 0.05 or 0.01.
D
and
d proposed furthher research
5.3 Discussion
Figure 12. Boxplots of diameteer d1 reveal no outliers.
o
Kolmogorov--Smirnov tesst was used
d to check the
difference in
n the means of
o results obttained in the first
(measuremen
nts performeed on fixed measured oobject
using only C
CMM) and in the second measurement
m
ccycle
(measuremen
nt obtained by
y the system robot-CMM).. The
results for all four measured
m
geometrical vaalues
own in Table 33.
(diameters d1 to d3 and heiight H) are sho
d1
Mean
CMM only
Mean
CMM-robot
Variance
CMM only
Variance
CMM-robot
D CMM only
p-value
CMM only
D CMM-robo
ot
p-value
CMM-robot
t Stat
P(T<=t) 1-tail
t Critical 1-taiil
P(T<=t) 2-tail
t Critical 2-taiil
d2
d3
Alth
hough this exa
ample shows that it is posssible to use an
n
indu
ustrial robot to
o extend the m
manipulation capabilities off
a co
oordinate measuring machin
ne, some impo
ortant aspectss
shou
uld be consid
dered. The exxperiment pe
erformed had
d
som
me disadvantag
ges, which aree summarized
d below.
Disa
advantages th
hat could havve affected the accuracy off
resu
ults included:
1.
H
39.0484
mm
39.0278
mm
6.97E-08
mm
3.80E-05
mm
0.166
42.5436
mm
42.5194
mm
1.46E-07
mm
4.21E-05
mm
0.107
42.9408
mm
42.9214
mm
1.86E-07
1
mm
2.98E-05
2
mm
0.091
18.88075
mm
18.88238
mm
2.48E
E-07
mm
2.36E
E-05
mm
00.129
0.458
0.920
0.980
00.771
0.109
0.116
0.201
00.151
0.912
0.871
0.235
00.585
2.
3.
16.7275
18.5872
17.6842
-16.77088
4.94E-15
4.69E-16
1.43E-15
1
2.24E
E-15
2.4992 (α<0.01), 1.71
11 (α<0.05)
9.89E-15
9.38E-16
2.87E-15
2
4.48E
E-15
2.7997 (α<0.01) , 2.06
64 (α<0.05)
Table 3. The reesults of the staatistical Kolmogorov-Smirnov ttest
(significance leevel p < 0.05) an
nd t-Test: Two-S
Sample Assumin
ng
Unequal Variaances (α<0.01, Hypothesized
H
Mean Difference 00).
4.
As the comp
puted p-value is greater tha
an the significcance
level 0.05, w
we cannot rejeect the null hy
ypothesis H0 (the
sample follow
ws a Normal distribution).
d
Levene’s tesst confirmed
d that varian
nces in the two
observed casses are differen
nt. Therefore the
t Welch’s t--Test,
Two-Sample Assuming Unequal Variances, was
performed in
n order to cheeck the differe
ence in the m
means
of the resullts obtained in the first and the seccond
measuremen
nt cycles.
of the t-Test, shown in Ta
able 3, lead too the
The results o
conclusion that Hypo
othesis 2 (“Uncertainty
(
of
www.intechopen.com
5.
Conditions of university laaboratory
Better equip
pment, laborratory completeness and
d
application of highest m
measuring sttandards can
n
he quality off
provide bettter conditionss and thus th
measuremen
nt results.
The bonding CMM and rob
obot
ation of the rrobot with th
he CMM wass
The combina
achieved as described in
n this paperr because off
technical and
d construction
on capacity constraints. In
n
order to incre
ease stability aand measurin
ng precision off
the robot, it is possible tto use different designs off
e using compleex binding ele
ements, which
h
bearing table
can enable the CMM and roobot to bind ass a single unit..
Vibrations caused by thee robot’s insttability on itss
bearing table (light consstruction, high
h position off
g
robot’s graviity centre by z axis, wheels on bearing
base, etc.).
uction of bearin
ng table and a
Larger and heavier constru
stronger link
k to the groun
und would po
ossibly reducee
vibrations du
uring movemeent of movablle elements off
the robot or CMM.
C
The con
nfiguration wh
here the robott
was fixed to the
t ground, w
without a physiical connection
n
to the CMM
M granite tablee, drastically increased thee
system’s stabiility and reducced vibrations.
Limited reach of the roboot’s arm due to
t its position
n
relative to the CMM.
M and robott
By using a different deesign of CMM
nt type of robo
ot, it would bee
binding, or even a differen
possible to increase thee overlapping workspacee
i
thee
zones of the robot and thee CMM, thus increasing
h.
robot’s reach
Mechanical impacts on C
CMM which occur whilee
shifting posiition of meassuring probe in phase off
measuring ne
ew geometricaal feature (surrface).
The design of the CM
MM used can cause thee
appearance of certain m
mechanical im
mpacts when
n
changing me
easuring phasee. These impa
acts can causee
vibration inccrease in thee robot’s arm
m, particularly
y
when at full stretch.
s
Samir Lemes, Damir Strbac and Malik Cabaravdic: Using Industrial
Robots to Manipulate the Measured Object in CMM
7
Future research in this area should be performed with
different configuration, with a more robust robot chassis,
and with more positions examined. Another improvement
would be to synchronize the software for CMM
manipulation and the software for robot manipulation,
providing real automation of the measurement process.
–
A deeper and more detailed measurement uncertainty
analysis, using both Type A and Type B errors, and
taking into consideration correlation of influence factors,
should also be performed, in order to give a more general
foundation for testing the complex measurement systems.
7. Acknowledgements
6. Conclusion
The principal idea of this paper was to extend the
possibilities for automating the measurement process
with coordinate measuring machines. The obstacle most
often encountered with CMM measurements are
limitations of geometry, requiring more measurement
sequences in order to reach difficult places on the
measured object. It is possible to perform measurements
of such objects, but manual repositioning of the measured
object, including redefinition of the local coordinate
system, slows down the process. If an industrial robot is
used to manipulate the measured object, such a process
could be automated. The ultimate goal is to keep the
measurement uncertainty within allowable limits The
measurements of the dimensions of the measured object
were conducted by complex CMM-robot measuring
system, with movements performed between each single
measurement. These results were compared with the
results obtained by measuring the same object fixed in the
CMM. The measurement results in these two cases were
different; one of the reasons for this could be the slight
impacts and vibrations that were obvious during every
movement phase between measurements.
Although the obtained measurement results still have
great accuracy and precision, they do not meet the criteria
of the CMM’s prescribed measurement uncertainty.
It can be concluded that it is possible to conduct
measurements using complex CMM-robot measuring
systems, but the measurement results are dictated by the
measurement uncertainty of the least accurate component
of the system, which in this case was the industrial robot.
Significant differences and deviations in measurement
results can be confirmed by comparing obtained
measurement results with results measured on an object
with a different mass. This confirms the significant
influence of variation in the mass of the measured object
on the measurement uncertainty of the complex CMMrobot measuring system.
It can be argued that it could still be possible to confirm
hypothesis 2, assuming the fulfilment of certain
conditions such as:
8
Int. j. adv. robot. syst., 2013, Vol. 10, 281:2013
–
Different design of CMM and robot combination,
which would reduce impacts and vibrations
occurring in CMM operation;
Use of newer and more advanced generations of
robots with greater capacity, stiffness, accuracy,
repeatability, etc.
This research was supported in part by the Ministry for
Education and Science of the Federation of Bosnia and
Herzegovina.
8. References
[1] J. Santolaria, J.J. Aguilar, Kinematic Calibration of
Articulated Arm Coordinate Measuring Machines
and Robot Arms Using Passive and Active SelfCentering Probes and Multipose Optimization
Algorithm Based in Point and Length Constrains,
Robot Manipulators New Achievements, A. Lazinica
and H. Kawai (Ed.), ISBN: 978-953-307-090-2, InTech,
2010
[2] B. Lawford, Uncertainty analysis and quality
assurance for coordinate measuring system software,
Master thesis, University of Maryland, 2003
[3] C.Y. Fang, C.K. Sung, K.W. Lui, Measurement
uncertainty analysis of CMM with ISO GUM, ASPE
Proceedings, Norfolk, VA, pp 1758-1761, 2005
[4] G. Zhang, R. Vaele, T. Charton, B. Brochardt, R.J.
Hocke, Error compensation of coordinate measuring
machines, Annals of the CIRP, Vol. 34/1, pp 445-451,
1985
[5] P.M. Ferreira, C.R. Liu, An Analytical Quadratic
Model for the Geometric Error of a Machine Tool,
Journal of Manufacturing Systems, 5:1, pp 51-60,
1986
[6] N.A. Duffie, S.M. Yang, Generation of Parametric
Kinematic
Error-Correction
Function
from
Volumetric Error Measurement, Annals of the CIRP
Vol. 34/1/1985 pp 435-438, 1985
[7] L. Adams, Wrapper Ties Robot to CMM, Quality, Vol.
41, Issue 6, pp 22, 2003
[8] H.N. Hansen, P. Bariani, L. De Chiffre, Modelling and
Measurement Uncertainty Estimation for Integrated
AFM-CMM Instrument, CIRP Annals - Manuf.
Technology, Vol. 54, Issue 1, pp 531-534, 2005
[9] F. Aggogeri, G. Barbato, E. Modesto Barini, G. Genta,
R. Levi, Measurement uncertainty assessment of
Coordinate Measuring Machines by simulation and
planned experimentation, CIRP Journal of Manuf.
Science and Technology, Vol. 4/1, pp 51-56, 2011
[10] A. Weckenmann, M. Knauer, H. Kunzmann, The
Influence of Measurement Strategy on the
Uncertainty of CMM-Measurements, CIRP Annals Manufacturing Technology, Vol. 47, Issue 1, pp 451454, 1998
www.intechopen.com
[11] R.G. Wilhelm, R. Hocken, H. Schwenke, Task Specific
Uncertainty in Coordinate Measurement, CIRP
Annals - Manufacturing Technology, Vol. 50, Issue 2,
pp 553-563, 2001
[12] C.-X.J. Feng, A.L. Saal, J.G. Salsbury, A.R. Ness,
G.C.S. Lin, Design and analysis of experiments in
CMM measurement uncertainty study, Precision
Engineering, Vol. 31, Issue 2, pp 94-101, 2007
www.intechopen.com
[13] A. Piratelli-Filho, B. Di Giacomo, CMM uncertainty
analysis with factorial design, Precision Engineering,
Vol. 27, Issue 3, pp 283-288, 2003
[14] NIST Engineering Statistics Handbook (accessed
30.3.2013), http://www.itl.nist.gov/div898/handbook/
prc/section2/prc222.htm
Samir Lemes, Damir Strbac and Malik Cabaravdic: Using Industrial
Robots to Manipulate the Measured Object in CMM
9