Recursive local kinematic inversion with dynamic task-priority allocation

Abstract

A general method for local kinematic inversion of non-redundant and kinematically redundant robotic mechanisms is proposed. The mathematical background is a recursive scheme based on gradient projection through pseudoinverses. The additional task (in case of a kinematically redundant robotic mechanism) and/or the end-effector tasks are decomposed into single task components. Priority among these components is allocated dynamically rather than keeping it fixed, as in other schemes. This approach yields the advantage of "sacrificing" only the worst-conditioned task components. Further on, it is shown how the notation can be modified to damp single solution components in the neighbourhood of task and/or algorithmic singularities, in order to guarantee some bounded solution norm.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>