Learn ability of skilled motion for robot systems with internal model

Abstract EN

In this paper a simple learning control scheme using internal model for trajectory tracking of robot manipulators is presented.

The proposed learning control structure consists of an internal model based controller plus a feed forward learning control for linear time varying system.

The two degree of freedom control structure consists of an inner disturbance controller for the system and used an outer tracking controller.

Instead of a quadratic artificial potential of joint angles, this paper introduces a new type of a quasi – natural potential in robot dynamics, which induces a type of sinusoidal position feedback with saturation in servo – loops.

The proposed algorithm is used for improving the performance at the next trail on the basis of the previous operation data.

Examples are given to show the effectiveness of the proposed algorithm.

An analysis of the convergence proof together with simulation results are presented.

The robustness of the algorithm against error in initial settings is studied through computer simulation.

Simulation results show good performance for the proposed algorithm.