A Recursive Identification Controller for Trajectory Tracking

Abstract

A controller for robot trajectory tracking is developed. This controller uses on-line recursive identification techniques to track the model parameters which describe the plant. The identified model is then used to compute a feedforward signal such that the desired trajectory is followed. This paper derives the trajectory tracking accuracy for a general class of dynamic systems. It also derives an identification technique for systems with an asymmetrical dead-zone. Issues such as identifiability, forgetting factors, calibration, and numerical stability of the algorithms are investigated. An inertially decoupled direct drive manipulator is used as a test bed to evaluate the performance of the recursive identification controller.