Achieving user-defined accuracy with damped least-squares inverse kinematics

Damped least-squares inverse kinematics gives well-behaved joint solutions even in the neighbourhood of singular configurations; this is obtained by weighting the tracking accuracy with respect to the feasibility of the resulting joint motion. When the task to be executed demands for high accuracy in assigned task-space components while allowing lower tracking performance in the other directions, the usual measure of the tracking accuracy may give rise to unsatisfactory performance close to singular configurations. In this paper it is shown how to tune the damped least squares inverse kinematics in order to improve the accuracy of the solution with respect to prescribed task-space components. The technique has been implemented to control an industrial manipulator in experiments.<>