Robotic Symposium Presentation

1. 1
An overview of the various
kinematic models in both
parallel & serial robotic mechanisms
and which are best suited
for precision applications
Ilian Bonev, Ph.D.
Professor, École de technologie supérieure
Montreal, QC, Canada

2. 2
Background

3. û founded 41 years ago
û 10,600 students: 8,000 in B.Eng. (already technicians) and 2,600 graduate
û Each year, 25% of new engineers in Quebec are graduates of the ÉTS
û third largest engineering school in Canada
û 180 professors and lecturers, 417 other personnel
û $35 million for research (75% from industrial collaborations)
3
École de technologie supérieure

4. û professor at ÉTS since 2004
û expert on parallel robots
û Canada Research Chair in Precision Robotics
û head of the Control and Robotics Lab (CoRo)
û cofounder of Mecademic
Ilian Bonev
4

5. Industrial robotics undergraduate course
û 4 ABB IRB 1600
û 1 ABB IRB 120 (mobile)
û Conveyors, PLCs, etc.
5

6. Control and Robotics Laboratory (CoRo)
6

7. Control and Robotics Laboratory (CoRo)
7

8. Parallel robots developed in CoRo
Precision XY-Theta positioning table (PreXYT)
8

9. Parallel robots developed in CoRo
Medical robot for vascular ultrasound examination
(MedRUE)
9

10. Parallel robots developed in CoRo
Dextrous twin-arm robot (DexTAR)
10

11. Parallel robots developed in CoRo
Micropositioning agile rotary hexapod (MicARH)
11

12. Parallel robots developed in CoRo
Hexapod with trivial kinematics (Hexapteron)
12

13. Parallel robots developed in CoRo
13
Cable-driven omnidirectional loading simulator
(CabOLS)

14. Robot calibration
14
A new device and method for calibrating robots

15. 15
Types of Mechanisms

16. 16
Serial vs parallel

17. 17
Serial vs parallel mechanisms
û large workspace
û stiff
û fast
û precise

18. Serial mechanisms
18

19. Serial mechanisms
19

20. Serial mechanisms
20

21. Parallel mechanisms
21

22. Parallel mechanisms
22

23. 2
3
Robot Calibration

24. Precision (Repeatability) vs Accuracy
24

25. Calibration procedure
25
1. A new mathematical model of the robot is developed.
2. (Partial) pose measurements of the end-effector are taken.
3. The parameters of the new model are identified thru optimization.
4. A “fake target” is computed for each end-effector pose.

26. Calibration of DexTAR with a ball plate
26

27. 27
Calib. of an IRB 1600 with a laser tracker
Mean error STD Max
With cable robot
(825 measurements,
120 N, 10 Nm)
0.264 mm 0.137 mm 0.960 mm

28. Calib. of an IRB 120 w/ laser t. & C-Track
28
(1 mm = 0.039 in)