Model-based adaptive position and force control of robot manipulators

摘要

This thesis is primarily concerned with motion control of robot manipulators with emphasis placed on adaptive impedance control and relevant computational issues. The general approach taken in our studies is a model-based approach, that is, the algorithms developed is based on the dynamic model of the robot(s) involved.
\n
\nTwo computational formulations are derived for the evaluation of the so-called regressor dynamics of a robot manipulator, which has played a key role in the development of popular stable adaptive control algorithms for robot manipulators. The closed-form version of the formulations is based on the Lagrangian dynamics formulation while the recursive version is based on the Newton-Euler dynamics.
\n
\nAs an application of the regressor dynamics formulation, the popular Slotine-Li adaptive control algorithm is modified and then implemented on a PUMA 560 robot. Satisfactory computational efficiency of the regressor formulas, especially the recursive formula, has been demonstrated in our experimental implementations.
\n
\nTo extend adaptive position control algorithms to force control, the concept of target impedance reference trajectory is introduced which makes it possible to inject two stable adaptive position control algorithms into Hogan's conventional impedance control. These two adaptive impedance control algorithms have been shown stable. Simulation and real-time implementation of the algorithms on a PUMA 560 robot are reported.
\n
\nThe last part of the thesis conducts a study on optimal load distribution and coordination of multiple robots. Optimal load distribution schemes using a p-norm type optimization approach are proposed. The algorithms are then adopted to dynamically link the two-level controllers in a proposed coordination framework. Simulation results are presented to show the performance of the proposed structure.