Robot Joint Acceleration Identification

Capture robot joint acceleration through experimental results.

Prerequisites

• Robot Motion Program Driver: ABB, FANUC, Motoman
• Tesseract: Collision Checking Library

Algorithm

We assume the robot is placed horizontally with base z-axis pointing upward along gravity. Since the arm inertia is most heavily influenced by the shoulder and elbow joints (joints 2 and 3), we assume that the spherical wrist joints (joint 4, 5, and 6) have constant joint acceleration limits, while joints 1, 2 and 3 acceleration limits depend on $(q_2,q_3)$, the shoulder and elbow joints, and joint 2 and 3 acceleration limits also depend on the direction of vertical motion due to gravity.

Results

For Joint 4, 5, and 6, acceleration is stored as constants. For the remaining, a dictionary is used with $(q_2,q_3)$ as keys and $(\ddot{q_1},\ddot{q_2},\ddot{q_3})$ are the value.

ABB6640	1	2	3	4	5	6
Velocity ($rad/s$)	1.745	1.571	1.571	3.316	2.443	3.316
Acceleration ($rad/s^2$)	*	*	*	42.5	36.8	50.5

ABB1200	1	2	3	4	5	6
Velocity ($rad/s$)	5.027	4.189	5.184	6.981	7.069	10.472
Acceleration ($rad/s^2$)	*	*	*	108.2	145.4	153.5

MA2010	1	2	3	4	5	6
Velocity ($rad/s$)	1.745	1.571	1.571	3.316	2.443	3.316
Acceleration ($rad/s^2$)	*	*	*	42.5	36.8	50.5

Usage

An example of Motoman and ABB is included as motoman.bat and abb.bat:

python capture_acc.py --robot-name=MA2010_A0 --robot-info-file=config/MA2010_A0_robot_default_config.yml --pulse2deg-file=config/MA2010_A0_pulse2deg.csv --displacement=0.05 --resolution=0.3 --q0-default=0.17 --robot-ip=192.168.1.31 --urdf-path=config/urdf/motoman_cell