DYNAMIC MODELING AND CONTROL OF ROBOTIC MANIPULATORS (ROBOTICS, CONTROL ENGINEERING, NONLINEAR CONTROL).

摘要

The objective of this dissertation is to advance the state-of-the-art in the dynamic modeling and feedback control of robotic manipulators with rigid links. The dissertation revolves around three complementary themes: robot dynamics, discrete-time dynamic robot modeling, and nonlinear feedback robot control. Each of these themes is addressed in turn.; Controller design for manipulators requires a fundamental physical understanding of the properties and structure of robot dynamics. Robot dynamics are embedded in the mathematical foundations of classical mechanics to introduce novel physical interpretations and structural characteristics of the closed-form dynamic robot model. This novel approach reinforces the need to integrate the mechanical and controller designs of robotic manipulators.; An inherently discrete-time dynamic robot model, which is compact and designed for computer-control of robots, is then introduced. The model guarantees conservation of energy (and momentum, if appropriate) over each sampling period. This discrete-time dynamic robot model is acceleration-free and is attractive for both forward and inverse dynamics applications in trajectory planning and dynamic feedback control.; A review of the fundamental principles of nonlinear feedback robot control algorithms leads to an outline of the practical problems introduced by modeling inaccuracies, unmodeled dynamics and parameter errors. The (alpha)-computed-torque nonlinear feedback control algorithm, which is robust in the presence of these modeling errors, is then introduced.; The aforementioned discrete-time dynamic robot model is used to (i) redesign the conventional computed-torque algorithm for computer implementation, and (ii) design robust discrete-time (accelerometer-free) computed-torque nonlinear feedback control algorithms. The contributions of the dissertation are illustrated through simulation experiments with the three degree-of-freedom positioning system of an articulated robot. The simulation experiments demonstrate the efficacy and applicability of the discrete robot model and robust nonlinear feedback control algorithms.