A near output feedback controller for a general class of robot manipulator systems

Abstract

In this paper, we propose a robust near output feedback (which refers to the fact that the control scheme uses dynamic feedback of the output but requires knowledge of the initial conditions of other states) link position tracking controller for a class of robotic systems, which includes the rigid-link flexible-joint (RLFJ) robot manipulator and the rigid-link electrically-driven (RLED) robot. The method uses a bank of dynamic filters, with the link position tracking error as the input, to generate surrogates for the unmeasurable system states. One of these generated filter states is then used in the feedback control strategy. The method is distinguished from previous results by the limited amount of modelling information required, the utilization of only link position measurements and knowledge of the initial conditions, and the overall simplicity of the controller. Even though the approach only requires link position measurements and knowledge of initial conditions, the controller exhibits good robustness properties while ensuring semiglobal uniformly ultimately bounded link position tracking.