Experimental comparison of model-based and model-free output feedback control system for robot manipulators

Abstract

In this paper, we design and implement model-free linear and model-based nonlinear adaptive output feedback method for robot manipulators. The model-free design uses only proportional and derivative (PD) error terms for trajectory tracking control of nonlinear robot manipulators. The design is very simple in the sense that it does not require a priori knowledge of the system dynamics. The model-based output feedback method combines PD controller terms with nonlinear adaptive term to cope with uncertain parametric uncertainty. The unknown velocity signals for two output feedback method are generated by the output of the model-free linear observer. Using asymptotic analysis, tracking error bounds for both output feedback design are shown to be bounded and their bounds can be made closed to the bound obtained with state feedback design by using small value of observer design parameters. Finally, we experimentally compare both method on a 3-DOF Phantom TM robot manipulator. © 2011 Springer-Verlag.