Modelling and Adaptive Control of a Mechanical Manipulator

Abstract

This paper implements and evaluates the adaptive control algorithm recently proposed by Horowitz and Tomizuka. A dynamic model of the Rhino XR-2 manipulator arm is developed using bond graph techniques. The manipulator and the control algorithm are simulated. The manipulator is then adaptively controlled. The study offers valuable insight into the practical problems associated with actual implementation of adaptive control algorithms and fully describes one of the few case studies of adaptive control in a working physical system. The physical problems directly addressed are disturbances due to Coulomb friction, gravitational loading, and actuator saturation.