A novel method for computing self-motion manifolds

Abstract

For kinematically redundant robots, self-motion manifolds that represent all the inverse kinematic solutions for a given end-effector location play a significant role in global motion planning. However, the efficient computation of self-motion manifolds, especially the computation of high-dimensional manifolds, is still a problem that needs to be solved. In this paper, the grid elements modeling criteria are formulated, and the specific modeling method of a grid element in the configuration space is developed. Based on the idea of cellular automata, the problem of computing self-motion manifolds is transformed into a dynamic model searching problem, and an evolution strategy is proposed to realize the efficient computation of self-motion manifolds. Finally, two examples are used to verify the effectiveness of the proposed method in computing high-dimensional self-motion manifolds. The results show that compared to the conventional methods, the proposed method can correctly identify all the self-motion manifolds with 94% reduced computational time on average.