Dataset link https://github.com/Lizhibing1490183152/RobotCali
A. Experimental setup
B. The critical components of Experimental setup
(1) An ABB IRB120 industrial robot ;
(2) A drawstring displacement sensor;
(3)A drawstring displacement indicator.
C. Dataset
- Manually teach the position of 1042 points in the robot workspace, the sampling points should cover the robot workspace, and their positions are measured by the drawstring displacement sensor.
- The drawstring displacement indicator displays the length of cable for the measuring points.
- The robot joint angles and position coordinate information can be obtained on the teaching pendant.
- We substitute the collected data and the robot kinematic parameters provided by the manufacturer into the calibration method for calculation to optimize the error.
D. The data format of RobotCali
The data file is excel xlsx format, each of data contains the robot joint angles, cable length and position coordinate information.
- x, y and z are the coordinate of the end-effector's position on the robot teaching pendant.
- q1 ~ q6 are the joint angles of the robot.
- L is the cable length of the measuring points.
E. The robot and measuring instrument specifications
Robot type: ABB irb120 industrial robot
Drawstring displacement sensor: HY 150-2000
Drawstring displacement indicator type: HY 9648
ABB IRB120 industrial robot specification:
Item | Specification |
Repeat positioning accuracy | 0.01mm |
Working radius | 580mm |
Weight | 25kg |
Load capacity | 3kg |
Height | 700mm |
Specification of the drawstring displacement sensor:
Item | Specification |
Signal output type | Digital signal |
Voltage | DC 5-24V |
Measuring range | 2000mm |
Maximum speed | 1000mm/s |
Expansion force | 5N |
Linear | 0.05% FS |
Resolving power | 0.004mm |
Working temperature | -25℃~+85℃ |
Specification of the drawstring displacement indicator:
Item | Specification |
Voltage | AC220V 50Hz |
Input resistance | 5K |
Pulse frequency | ≤500K |
Zoom in range | 0.000~999.999 |
Display range | -199999~999999 |
Pulse width | 2us |
Working temperature | -10℃~50℃ |
Rbotcali download instructions:
joint i | αi /◦ | ai /mm | di /mm | θi /◦ |
1 | -90 | 0 | 290 | 0 |
2 | 0 | 270 | 0 | -90 |
3 | -90 | 70 | 0 | 0 |
4 | 90 | 0 | 302 | 0 |
5 | -90 | 0 | 0 | 0 |
6 | 0 | 0 | 72 | 0 |
where ai, di, θi and αi are the link length, link offset, joint angle, link twist angle, respectively.
F. The position calculation of robot end-effector
(1)Robot kinematic model
T is link transformation matrix., cθ is cosθ,sθ is sinθ.
(2)Calculation of theoretical cable length
where is the measuring length of cable ,is the theoretical length of cable,is the theoretical position of robot end-effector,is measuring position of robot end-effector ,is he coordinate of ground fixed point.
We can optimize the error according to the difference between the theoretical calculation of the cable length and the measurement of the cable length. LM algorithm, extended Kalman filter, particle filter, PSO algorithm, genetic algorithm and neural networks can be used to optimize the robot position errors. Readers can choose their own optimization algorithms to optimize the robot position errors for publishing extensively excellent papers.
G. Reference
(1)Z. Li, S. Li, and X. Luo, “An Overview of Calibration Technology of Industrial Robots,” IEEE/CAA J. Autom. Sinica, vol. 8, no. 1, pp. 23-36, Jan. 2021.
(2)Z. Li, S. Li, and X. Luo, “Data-driven Industrial Robot Arm Calibration: A Machine Learning Perspective,” 18th IEEE International Conference on Networking, Sensing and Control, 2021.
(3)Z. Li, S. Li, O. O. Bamasag, A. Alhothali and X. Luo, “Diversified Regularization Enhanced Training for Effective Manipulator Calibration,” IEEE Trans. Neural Networks and Learning Systems, DOI: 10.1109/TNNLS.2022.3153039.