An Experimental Evaluation of Reinforcement Learning for Gain Scheduling

Most conventional approaches of force control for surface following operations require fine tuning of the feedback gain to be successful. The optimal feedback gain values of the force control loop are either analytically derived based on the geometrical model of the surface or determined empirically. This paper presents an experimental investigation of using reinforcement learning techniques to generate a gain schedule for an unknown surface. The result is compared with fixed and constant gain values.

[1]  Karim Djouani,et al.  Neuro-fuzzy based approach for hybrid force/position robot control , 2002, 2002 IEEE International Conference on Industrial Technology, 2002. IEEE ICIT '02..

[2]  Masaki Yamakita,et al.  Gain scheduled control of robot manipulators for contact tasks on uncertain flexible objects , 1996, 1996 IEEE International Conference on Systems, Man and Cybernetics. Information Intelligence and Systems (Cat. No.96CH35929).

[3]  R. Bellman.  Dynamic programming. , 1957, Science.

[4]  M. A. Jarrah,et al.  Position control of a robot manipulator using continuous gain scheduling , 1999, Proceedings 1999 IEEE International Conference on Robotics and Automation (Cat. No.99CH36288C).

[5]  John J. Craig,et al.  Hybrid position/force control of manipulators , 1981 .

[6]  Richard S. Sutton,et al.  Learning to predict by the methods of temporal differences , 1988, Machine Learning.

[7]  Richard S. Sutton,et al.  Introduction to Reinforcement Learning , 1998 .

[8]  Matthew T. Mason,et al.  Compliance and Force Control for Computer Controlled Manipulators , 1981, IEEE Transactions on Systems, Man, and Cybernetics.

[9]  D. E. Whitney,et al.  Historical Perspective and State of the Art in Robot Force Control , 1987 .

[10]  Ganwen Zeng,et al.  An overview of robot force control , 1997, Robotica.

[11]  Andrew G. Barto,et al.  Learning to Act Using Real-Time Dynamic Programming , 1995, Artif. Intell..

[12]  William Towsend,et al.  The Effect of Transmission Design on Force-Controlled Manipulator Performance. , 1988 .

[13]  Peter Dayan,et al.  Q-learning , 1992, Machine Learning.

[14]  Shuzhi Sam Ge,et al.  Adaptive NN impedance control of constrained mechanical systems , 2002, Proceedings of the 2002 American Control Conference (IEEE Cat. No.CH37301).

[15]  Kenneth Salisbury,et al.  Whole arm manipulation , 1988 .