Evaluation System(RPES) Software was developed with Visual Basic satisfying the need of Hyundai. Motor Company. The developed system was implemented ...
Development of a Robot Performance Evaluation System Using Leica LTD 500 Laser Tracker
Mi-Kyung Kim, Cheon-Seok Yoon (Graduate School, UOU), Hee-Jun Kang(UOU) School of Electrical Engineering, University of Ulsan San-29 Moogu-2 Dong, Namgu, Ulsan 680-749, Korea ABSTRACT A Robot Performance Evaluation System(RPES) with the laser tracker Leica LTD 500 was developed according to the ISO 9283 robot performance criteria. The developed system is set up a test robot to continuously move the prescribed cyclic trajectories without a human intervention and the laser tracker to simultaneously measure the robot’s movement. And then, the system automatically extracts the required data from the tremendous measured data, and compute the various performance criteria which represents the present state of the test robot’s performance. This paper explains how ISO 9283 robot performance criteria was used for the developed system, and suggests a automatic data extraction algorithm from the mass of measured data. And also, a user-friendly Robot Performance Evaluation System(RPES) Software was developed with Visual Basic satisfying the need of Hyundai Motor Company. The developed system was implemented on NACHI 8608 AM 11 robot. The resulted output shows the effectiveness of the developed system. Key Words : Robot Performance Evaluation System(RPES), RPES Software, ISO 9283 Robot Performance Criteria
1. INTRODUCTION Production of various products and upgrading their qualities have been accelerated by recent development of electronic and mechanic technology. And also manufacturing technique of assembly automation and checkup automation etc. have steadily grown both quantitatively and qualitatively. Especially, industrial robots play important roles in automation production lines that aim for high quality product and its quality regularities. For this purpose, the standard way of evaluating the performance of industrial robots must be necessary. Hereupon, ISO have made a standard of robot performance criteria in 1990, and also given out a standard for the way of evaluating the performance criteria of industrial robots in 1995.[1,2] This research aims for development of a
robotic system that evaluates the performance of a robot according to ISO regulation. The developed system is set up a test robot to continuously move the prescribed cyclic trajectories without a human intervention and the laser tracker to simultaneously measure the robot’s movement. And then, the system automatically extracts the required data from the tremendous measured data, and compute the various performance criteria which represents the present state of the test robot’s performance. This paper explains how ISO 9283 robot performance criteria was used for the developed system, and suggests a automatic data extraction algorithm from the mass of measured data. And also, a user-friendly Robot Performance Evaluation System(RPES) Software was developed with Visual Basic satisfying the need of Hyundai Motor Company. The developed system was implemented on NACHI 8608 AM 11 robot[3]. The resulted output shows the
effectiveness of the developed system. 2. The Configuration of Robot Performance Evaluation System Robot performance evaluation system (RPES) consists of a target robot(currently, NACHI 8608 AM 11), LTD 500 Laser Tracker (Leica Geosystems Ltd . Switzerland)[4] shown in Fig 1. LTD laser tracker sensor measures the 3 dimensional position based on the designated reference coordinate frame. By using the principle of laser interference. It emits the laser to the designated reflector and receives the its returning laser beam. The way of measuring a target position of robot tip is explains as follows; 1) a test robot is placed in a precisely manufactured plate which has four reflector mounts. 2) After a reflector was placed at reflector mounts, LTD 500 sensor measures those 4 positions and finds out the location of the robot base frame. 3) Then, it allows the robot base frame to the reference frame for the future measurement. 4) While the robot moves, the LTD laser sensor measures the target position at which the reflector is mounted. The robot moves along the predefined cyclic trajectories downloaded to robot controller through off-line programming using ROBCAD (Tecnomatix Technologies Ltd . Israel) Next sections explains how the predefined trajectories are constructed for the evaluation of robot’s performance according to ISO 9283. 2.1. Position of cube from working space According to ISO 9283, a measurement cube must be virtually placed within robot’s workspace, satisfying the following conditions. - The cube has to be placed at the space where the robot is most frequently used within its workspace - The cube must have its possible maximum dimension and it has the sides that should be parallel to the robot base frames. The obtained cube from the conditions explained above is shown in Fig 1 for our target robot. Its dimension is 1200*1200*1200mm and its corner positions are given in Table 1. 2.2. Poses that are tested According to ISO 9283, five order pose (P1 ~ P5)s are set in diagonal line of measurement
plane (opposite angle plane C4 - C1 - C6-C7 of cubes) C2 C1 C3 C4 C6
C5
C7 C8
Fig. Robot Performance Evaluation System Table 1 Test Cube Coner Pos
x
y
z
Cube C1
600
2300
2000
Cube C2
600
1100
2000
Cube C3
-600
1100
2000
Cube C4
-600
2300
2000
Cube C5
600
2300
800
Cube C6
600
1100
800
Cube C7 Cube C8
-600 -600
1100 2300
800 800
P1 is a intersection point of the diagonal lines and center of cube. P2, P3, P4, P5 are set at the points which have distances (10±2)% of the length of the diagonal line from both end of the diagonal lines The positions P2, P3, P4, P5 are set distant from the center of cube C4 by each 100mm in x direction, y direction, z direction of the reference coordinate frame shown in Fig.2
P3 P
P2 P
P1 P P5
P
P4
Fig. 2 Command Positions
3. Robot Performance Criteria ISO 9283 suggests robot performance criteria which evaluate the robot pose characteristics and the robot path characteristics. According to ISO 9283, the developed Robot Performance Evaluation System (RPES) covers the following performance criteria:. Position accuracy and repeat accuracy Position Overshoot and stabilization time Multi direction position accuracy deflection Distance accuracy and Distance repeat accuracy Mini pose time Next sections will briefly explains what they are and how they are incorporated in RPES.
The cyclic trajectories starts from P1 to P2, P3, P4, P5 and back to P1, which completes 1 cyclic trajectory. The robot moves n times repeatedly, and the robot stops during 0.5sec at the commanded position such as P1-P5. This temporary stops gives us to set up an algorithm to identify the poisition value from the the tremendous measured data by LTD sensor. Think the data is measured at every 0.01 sec. From those data, the target data must be extracted. For the data extraction algorithm suggested in this paper, a moving average velocity are defined as v (k ) =
1 k
k
∑ v(i) ,
(3)
i =1
where, instant velocity v(i) = v x2 (i) + v 2y (i) + v z2 (i) , 3.1 Position Accuracy and Repeatibility Pose Accuracy(PA) is expressed as the standard deviatrion with "average of arrival poses" in the same direction as "command pose". Pose: APp =
(x − xc )2 + ( y − yc )2 + (z − zc )2
where, x =
1 n
n
∑x
, y=
j
j =1
1 n
n
∑y j =1
j
, z=
(1) 1 n
j
.
j =1
RPl = l + 3S l ,
lj =
1 n
(2) n
∑l
j
,
j =1
(x j − x )2 + (y j − y )2 + (z j − z )2 , n
∑ (l − l )
2
j
Sl =
j =1
n −1
v y (i ) =
p x (i ) − p x (i − 1) , t (i ) − t (i − 1)
p y (i ) − p y (i − 1)
, t (i ) − t (i − 1) p (i ) − p z (i − 1) v z (i ) = z . t (i ) − t (i − 1)
n
∑z
Pose Repeatibility(PT) is expressed as the worst difference between the each trial and its average of arrival when robot repeatedly moves towards the same point in the same direction of approach. The detailed equation is shown as where, l =
v x (i ) =
.
3.1.1 Evaluation Method for PA and PR In this section, the process of evaluating Poses Accuracy and Pose Repeatibility is explained with Nachi8608 robot. Robot moves to the command poses P1, P2, P3, P4, P5, repeatedly n timees with the predetermined maximum load of 100kg and maximum speed of 2300 mm/sec. LTD 500 continuously measures the target points with sampling time of 0.01 sec.
Fig.3 shows the instant velocity graph for the first 3 cyclic trajectories. While first area (0 ~ 4 seconds) is waiting in P1 in Fig 3, it is area that begin measurement. First acceleration area (4 ~ 7 seconds) is area that move from P1 to command pose P2 Deceleration area is area that approach by command pose P2 All - processing process of performance evaluation algorithm is that arrival pose to draw automatic justly in this deceleration area "Smallest velocity area" find and draw arrival pose P2 in smallest velocity area Second acceleration area (7 ~ 9 seconds) is area that move from P2 to order pose P3, and deceleration section is area that approach by order pose P3. RPES extracts arrival pose P3 in deceleration area. Third deceleration area (9 ~ 11 seconds) extracts arrival pose P4. and Fourth deceleration area (11 ~ 13 seconds) extracts arrival pose P5. and Fifth deceleration area (13 ~ 15 seconds) extracts arrival pose P1
Fig. 3 the first three cycles of Command Pose P1-P2-P3-P4-P5
Fig. 4 the Moving Average of the velocity and the Average of the velocity of the Moving Horizon Area that 2nd cycle's the first arrival pose P2 is extracted is 14.5 ~ 17.5 second, and it is same with Fig 4. As explain in front, acceleration area is area that move from P1 to order pose P2, and deceleration area is area that approach by order pose P2. Arrival pose can extract in place that the velocity is smallest in deceleration area. In this case, receive best robot repeat accuracy by algorithm drawing best arrival pose. Save "Smallest area" to avoid case of front. and use way to choose the section's free point M j = {p(j-M) , p(j-(M-1 )) , p(j-(M-2 )) , … , p(j-1 ) , p(j) , p(j + 1 ) , … , p(j + (M-2 )) , p(j + (M-1 )) , p(j + M)}
,
(4)
j = M p , 2 M p , 3M p , …
If Moving Horizon defines to get smallest area, it is as following When measurement space (sampling time) is 0.01 secs, it becomes moving area (m=10) of 0.2 secs space that consist of 21 measurement point.
If moving cycle is 10, (mp = 10), (current) measurement point of last 10 become next measurement point of first 10 Calculate sum (sum of the speed in measurement point of a 2m+1 dog) of the velocity each moving area in given deceleration area. If sum of the moving area velocity of this is least, Moving area becomes for smallest smallest velocity area RPES extracts center point (datum point, p (j) in transfer section) in arrival pose among measurement point of 2m+1 of smallest velocity area 3.2 Position Overshoot and Stabilization Time In this section, Position Overshoot(PO) and Stabillization Time(PST) will be explained as another important robot peformance criteria. Position Overshoot is used for representing smooth and correct stop ability around arrival pose (position). Position Overshoot is defined as maximum distance of arrival pose between distances with measurement points that escape limit area Here, m are number of measurement point after end of robot enters by limit band by first Regulation limit is defined into single value that is defined by position repeat accuracy or is described by manufacturer POS = max POS j ,
(5)
POS j = max Dij , max Dij > ±ÔÁ¤ÇÑ° , = 0 , max Dij < ±ÔÁ¤ÇÑ° max Dij = max
(xij − x j )2 + (yij − y j )2 + (zij − z j )2 ,
i = 1,2,..., m.
Position Stabilization Time(PST) is used for representing how fast the motion robot is stabilized around the command pose.Position Stabilization Time is defined as an elapsed time to ( t mj ) moment remain within limit band from ( t1 j ) moment end of robot entered by first by limit band RPES presents average value that repeat n times of position stabilization time PST =
1 n
n
∑ PST
j
,
j =1
PST j = t mj − tij , i = 1,2,..., m .
(6)
Here, m are number of measurement point after end of robot enters in limit band by first 3.2.1 Evaluation Method for PO and PST Repeat accuracy to yield position Overshoot and stabilization time can define position repeat accuracy (in this case robot of same kind is presented into single cost that is equal.) that robot manufacture four presents or position repeat accuracy (repeat accuracy of target robot) that is presented by robot performance estimation by regulation limit. Position repeat accuracy of NACHI 8608 that manufacturer presents is 0.5mm End of robot appeared to approach by arrival pose P1 with limit band in Fig 5 End of robot enters and does not escape limit band to 46 seconds' arrival pose by limit band again at 45.23 seconds while escaped limit band at 45.22 seconds after entered by limit band at 45.1 seconds in this example Position Overshoot about arrival pose P1 becomes maximum value 0.501762mm in 45.22 ~ 45.23 second Position stabilization time is 0.13 that 45.23 seconds and difference of 45.1 seconds When define limit band by P1's position repeat accuracy, Fig 5 is result. In this case, being evaluated that end of robot approaches in arrival pose, position Overshoot becomes 0mm and position stabilization time gets into 0 seconds.
seconds. And End of robot entered by limit band again at 66.27 seconds. And End of robot did not escape limit band to arrival pose of 66.6 seconds. Position Overshoot about arrival pose P1 is 0.223278mm in 66.22 ~ 66.27 second, Position stabilization time becomes difference (0.11 seconds) of 66.27 seconds and 66.16 seconds.
Fig. 6 Position Overshoot – Case2 (Left) POS for the limit 0.5mm, (Right) POS for the limit 0.2mm
3.3 Multi Direction Position Accuracy When visit n times in same order pose of from three right angle direction, Multi direction position accuracy deflection is deflection between each other average arrival pose. vAPp = max
(xh − xk )2 + ( y h − yk )2 + (z h − z k )2
, (7)
h, k = 1,2,3
Fig. 5 Position Overshoot – Case1 (Left) POS for the limit 0.5mm, (Right) POS for the limit 0.2mm Though other several cases exist, Fig . 6 is the example Also, it is approximation area (65.5 ~ 67 seconds) of arrival pose P1 End of robot approaches about regulation limit 0.5mm in this example but can know that position Overshoot happens about (left picture), regulation limit 0.2mm End of robot escaped limit band at 66.22 seconds after enter by limit band at 66.16
Fig. 7 Multi-directional Pose Accuracy (Upper graphs) vAP for P1, (Middle graphs) vAP for P2, (Lower graphs) vAP for P4.
3.4 Distance Accuracy and Distance Repeat Accuracy Distance accuracy is position set and deflection of direction between average of arrival distance and order distance.
AD p = D − Dc
Acknowledgement
Distance repeat accuracy is approximation between several arrival distance that is repeated for direction such as n times about same order distance. n
∑ (D RD = ±3
j
−D
)2 REFERENCES
j =1
n −1
3.5 Minimum Pose Time Minimum pose time is difference of start time and end time of arrival pose ISO is defining Minimum pose time that calculates the average value after 3th repeat MPT =
1 3
This work was supported in part by University of Ulsan(2004) and also by Hyundai Motor Company.
3
∑ MPT
j
j =1
Fig. 8 (Left) Distance Accuracy and Repeatability, (Right) Minimum Posing Time.
4. Conclusion In this paper, A Robot Performance Evaluation System(RPES) with the laser tracker Leica LTD 500 was developed according to the ISO 9283 robot performance criteria. The developed system is set up a test robot to continuously move the prescribed cyclic trajectories without a human intervention and the laser tracker to simultaneously measure the robot’s movement. And then, the system automatically extracts the required data from the tremendous measured data, and compute the various performance criteria which represents the present state of the test robot’s performance. This paper explains how ISO 9283 robot performance criteria was used for the developed system, and suggests an automatic data extraction algorithm from the mass of measured data. And also, a user-friendly Robot Performance Evaluation System(RPES) Software was developed with Visual Basic satisfying the need of Hyundai Motor Company. The developed system was implemented on NACHI 8608 AM 11 robot. The resulted output shows the effectiveness of the developed system.
1. KS B 7082: 1999 (ISO 9283: 1998) Robot Peformance Criteria, Association of Korean Standard, 1999. 2. KS B ISO/TR 13309: 2002 Industrial Robot – The experimental methodology for robot performance evaluation based on ISO 9283, Association of Korean Standard,, 2002. 3. Nachi Robot 8000 AM10, AM11 Controller Maintenance Manual, Nachi-Fujikoshi Corporation. 4. Axyz Training Manual for Trackers, Version 1.4.0, Leica Geosystems Ltd, 2000. 5. eM - Workplace Training (ROB 100), Tecnomatix Technologies Ltd., 2000. 6. Bryan Greenway, “Robot accuracy,” Industrial Robot: An International Journal, Vol. 27, No. 4, pp. 257 - 265, 2000..
