Adaptive Cartesian Control of Redundant Manipulators

Abstract

Adaptve control of robotic manipulators can effectively utilize the ever improving capabilities of computers and mechanical systems. It combines high-speed and high-precision and provides consistent performance in the face of large variations in loads and tasks. Previously, a simple, globally tracking-convergent, direct adaptive controller has been developed and experimentally verified. This paper points out some recent findings regarding the effective implementation directly in the cartesian work-space of the end-effector. These include a filter-like procedure to avoid differentiation of the Jacobian pseudo-inverse. A passivity interpretation of the controller and its application to constrained motion is also provided and finally some experimental results of cartesian control are shown.