An efficient algorithm for dexterous manipulation planning

This paper addresses the dexterous manipulation planning problem of 3D rigid objects by a multi-fingered hand. We present a general motion planning algorithm capable to automatically generate specific stable grasps allowing a multifingered hand to manipulate rigid objects. It is also capable to address continuous sets of stable grasps, rather than sampling one generally assumed by the previous planners. The algorithm relies on a topological property that characterizes the existence of solutions in the subspace of configurations where the hand grasps the object with four fingers [13]. This property leads to reduce the problem by structuring the search-space. Experiments conducted with the planner demonstrate its efficiency to solve complex dexterous manipulation problems. Keywords— Dexterous manipulation, stable grasps, motion planning, probabilistic roadmap methods.