Planning motions compliant to complex contact states

Many robotic tasks require compliant motions, but planning such motions pos es special challenges not present in collision-free motion planning. One ch allenge is how to achieve exactness, that is, how to make sure that a plann ed path is exactly compliant to a desired contact state, especially when th e configuration manifold of such a contact state is hard to describe analyt ically due to high geometrical complexity and/or high dimensionality. The a uthors tackle the problem with a hybrid approach of direct exploitation of contact constraints and randomized planning. They particularly focus on pla nning motion that maintains certain contact state or contact formation (CF) , called a CF-compliant motion, because a general compliant motion is a seq uence of such CF-compliant motions with respect to different CFs This paper describes a randomized planner for planning CF-compliant motion between tw o arbitrary polyhedral solids, extending the probabilistic roadmap paradigm for planning collision-free motion to the space of contact configurations. Key to this approach is a novel sampling strategy to generate random CF-co mpliant configurations. The authors also present and discuss examples of sa mpling and planning results.