LEARNING CONTROL OF MOTION AND FORCE FOR CONSTRAINED ROBOTIC MANIPULATORS

摘要

Industrial robotic manipulators are usually used to perform repetitive operations and in many situations have to operate in constrained environments. Owing to the nonlinear and nonclassical nature of the dynamic model of constrained robots, it is difficult to design a controller to satisfy certain performance specifications for accurate tracking. Furthermore, in the case of repetitive operations, whatever errors may exist in following a trajectory will be repeated in the subsequent movements. In this paper, the authors consider constrained robots and design a learning control algorithm to reduce both motion and force errors iteratively. The proposed controller is able to improve performance based on the previous operation as the action is repeated. Simulation results of a constrained cylindrical robot are presented to illustrate the proposed controller.