Resolution of robot redundancy in the Cartesian space by criteria optimization

A computationally efficient control law for a redundant robot in the Cartesian space is presented. This approach uses a decomposition of the joint acceleration into primary and secondary joint acceleration. The joint acceleration solution is searched in order to compute the dynamic control law which can achieve the desired task in the Cartesian space with smaller tracking errors, little calculation, and good system behavior. The extra degrees of freedom are used to optimize a quadratic criterion by a direct or a gradient method. The results are compared to the dynamic control proposed by Hsu and colleagues in `Dynamic control of redundant manipulator´, Proc. IEEE Conf. on Robotics and Automation, 1988, p.183-7 for a 3-joint planar robot. Similar results are obtained with less calculation