ADVANCED CONTROLLERS FOR ROBOTIC MANIPULATORS

Abstract

The aim of the work reported in this thesis is to develop engineering-oriented algorithms for the high-performance control of robotic manipulators. The measure taken for performance improvement is twofold: (i). more complete model of manipulator systems with less unmodelled dynamics is adopted in order to achieve high-speed high-precision manipulation; (ii). additional structure for uncertainty reduction and disturbance attenuation is appended to conventional model-based control loops in order to enhance the robustness of model-based control systems. The major contributions of this thesis can be summarized as follows: • Handcrafted algorithms for the decoupling and linearization of manipulator systems with a dynamic model that includes both the mechanical dynamics of the links and the electrical dynamics of the motor actuators a.re developed. By using the proposed nonlinear transformations, the highly nonlinear and strongly cross-coupled electromechanical dynamics of the motor-manipulator systems can be decoupled and linearized without using advanced mathematical tools. And the developed algorithms a.re more compact and straightforward than other existing methods of its kind. Control of motor-manipulator systems is thus made easy with these simple algorithms. • Based on the physics of the cross-interacted electromechanical dynamics of motor-manipulator systems, algorithms for disturbance rejection and uncertainty attenuation of manipulator systems which are based on disturbance observation and feedforward compensation a.re developed. By using the proposed algorithms, not only can the possible external disturbances and uncertain dynamics acting at manipulator joint, be rejected and attenuated successfully, but the modelling difficulty and the complexity of control implementation can also be reduced significantly. • By exploiting the structural properties of the conventional model-following control, a simple scheme for the robustness enhancement of model-based control strategies is proposed. The scheme with a compact quantitative result of uncertainty reduction is an ideaI choice for the robustness enhancement of model-based control systems. It is also shown that the proposed scheme is general and can be easily added to existing control loops. Less structural modification to the existing control systems and less computation are required for its practical implementation. The proposed scheme is then used to enhance the robustness of the model-based control algorithms for robotic manipulators developed in this thesis. Experimental studies have been conducted to verify the performance of the algorithms reported in this thesis. The details of the thesis are specific to a prototype manipulator system, but the underlying principles also apply to many other dynamic plants. In developing the advanced controllers presented in this thesis, much 6ft has been devoted to reducing the mathematical complexity of the deviation presentation. Thus, these engineering-oriented algorithm can well be appreciated by application engineers. And it is believed that with the algorithms developed this thesis, control engineers can implement high-performance control systems the fast and easy way.