Stable inversion based output tracking control of robotic systems

Abstract

This thesis addresses stable inversion based output tracking control and its applications to robotic systems. It considers the non-causal invertibility (stable inversion) problem of control systems in its various aspects including properties of stable inverses and algorithms for constructing stable inverses. Then, the stable inversion approach is applied to solve a control problem of long-standing interest: output tracking control for non-minimum phase nonlinear systems. A minimum energy property of stable inverses is firstly established. The property claims that given any desired output trajectory, out of infinitely many possible inverse solutions, the one provided by the stable inversion process is the only one that has finite energy. Based on this property, a numerical procedure is developed to provide an efficient approach to construct stable inverses. Secondly, a new output tracking control design is developed. The design incorporates stable inverses and assumes a controller structure of feed-forward plus feedback. It achieves high precision tracking together with closed-loop stability. Furthermore, when system uncertainties are considered and assumed to satisfy the so-called matching conditions, a modified controller structure is presented and the corresponding robust tracking performance is discussed. Finally, the stable inversion based tracking control design is applied to three flexible robotic systems. The first study is output tracking control of a flexible-joint robot. The application demonstrates how the new design deals with the undesirable non-minimum phase property and achieves desired output tracking. The second application is tip trajectory tracking for a two-flexible-link manipulator. This thesis, for the first time, addresses the problem of stable tip trajectory tracking without any transient or steady-state errors for such non-minimum phase systems. In the third application, a new optimal motion control strategy for a flexible space robot is presented. The space robot system is assumed to consist of a two-link flexible manipulator attached to rigid space-craft. Optimality is in the sense that a performance index measured by maneuvering time, control effort, and structural vibrations is minimized while the interference from the manipulator to spacecraft is kept satisfactorily small. Studies on three applications demonstrate that the stable inversion based control design is very effective on output tracking for various robotic systems. This approach is expected to perform equivalently well for many other realistic non-minimum phase nonlinear systems.