Extended High-gain Observer Based Output Feedback Linearization of Robot Manipulator

摘要

Feedback linearization is a classic approach in robot manipulator control. In this paper, it is assumed that only joint position measurements are provided. One difficulty with feedback linearization is model uncertainty. We propose a technique to handle perturbation caused by parameter error or external disturbance. The robot system is treated as a perturbation of a nominal model, which can be linearized with state feedback control. An Extended High-Gain Observer (EHGO) is designed to estimate angular velocities and the perturbation. The perturbation is cancelled with the perturbation estimate. Velocity terms in the state feedback control law are replaced by the velocity estimates. We prove that the performance of the nominal linearized system can be recovered. Besides, with a high-gain observer fast enough, the tracking error converges to zero when the external disturbance is constant. Simulation results of a 2-DOF manipulator are included, featuring different aspects of this method.