Adaptive Control Strategies for Computer-Controlled Manipulators.

Abstract

The main problem in designing a controller for a manipulator stems from the complexity of the manipulator dynamics, which are typically described by a set of highly coupled nonlinear second order differential equations. In this thesis the overall manipulator control is improved by developing adaptive control strategies which yield high performance over a wide range of manipulator motions and payloads. The adaptive control strategies are based on perturbation equations in the vicinity of a desired trajectory. The overall controller is characterized by an open-loop feedforward component and a feedback component. The feedforward component computes the nominal torques which compensate for the interaction forces among the various joints and links. For the feedforward component, two different types of compensators are used: a partial feedforward compensator which compensates for the gravity effects of a manipulator, and a complete compensator which compensates for gravity and inertial effects. The feedback component determines the variational torques which reduce the position errors of the manipulator along a nominal trajectory. A recursive least squares identification scheme and optimal one-step adaptive control scheme are used to perform on-line parameter identification and control of the linearized perturbed system. The adaptive control strategies, based on feedforward compensation, reduce the manipulator control problem from a nonlinear control problem to a problem of controlling a linear system about a desired trajectory. A clear advantage of this adaptive control method is that since the nominal and variational torques can be computed separately and simultaneously, high performance real time control of the manipulator can be achieved. Extensive computer simulation studies of a three-joint PUMA robot arm are carried out to demonstrate the performance of the proposed adaptive control strategies. The relative advantages, limitations, and implementation of these proposed adaptive control strategies are discussed.