ENHANCEMENT OF COMPUTED-TORQUE CONTROL FOR ROBOT MANIPULATORS

Abstract

Robust control and adaptive control are the two major trends in the trajectory following control problem of robot manipulators. The attractive features of the robust control approach are that on-line computation is kept to a minimum and it has an inherent robustness to bounded disturbances. The advantages of the adaptive control approach are that the control implementation does not require a priori knowledge of unknown parameters in the robot dynamic model and the control performance improves with time. This dissertation focuses on the development of two robust control schemes and two adaptive control approaches for the trajectory tracking control of robot manipulators. All these control schemes are developed based on the basic structure of the computed-torque control (CTC) algorithm, which has a wide popularity in modern robot control theories. Hence, the core of the dissertation can also be interpreted as an enhancement of the CTC scheme for robot manipulator. Robustness was introduced to the robot CTC system by using the internal model control (IMC) approach. IMC is a robust control structure initially motivated for process control problems. This structure provides a simple yet explicit framework for the transparent analysis of control system performance, stability and robustness. However, to our best knowledge from a review of the literature, this effective control approach has not been widely applied for the control of a robot manipulator system so far. This dissertation makes an original attempt on this issue. A novel fixed-structure compensator was also designed to further improve the robustness of the enhanced CTC scheme incorporated with the IMC framework. The adaptive capability was introduced to the robot CTC system by using neural network methodologies. Two newly-developed NN adaptive components, based on the back-propagation (BP) scheme and the adaptive linear element (ADALINE) algorithm respectively, were successively implemented in the CTC system to enhance the control performance in face of uncertainty. The control performance of the conventional robot computed-torque controller and its four modified structures with improved adaptive ability and robustness are analyzed and compared. Simulation and experiment studies on a one-link and a two-link robot were used to verify the theoretical results.