Fast Motion on Cartesian Paths by Exploiting Robot Redundancy at the Acceleration Level

Abstract

Minimizing the transfer time along a given Cartesian path for redundant robots can be approached by separating the generation of a parametrized joint path associated to the Cartesian path from the exact minimization of motion time under kinematic/dynamic bounds along the obtained joint path. In this framework, we have multiple solutions depending on how redundancy is resolved at the kinematic level. We propose to achieve this at the acceleration level, by using weighted pseudoinversion, optimizing an inertia-related criterion, and including null space damping. Numerical results on a 3R planar arm and with the 7R KUKA LWR robot show consistently good behaviors and also improvements in comparison with related methods previously proposed in the literature.