Discretized Caging Set of Three Dimensional Arbitrary Rigid Bodies and Multifingered Hands

Abstract

This abstract tackles an open problem of robotic caging grasps, in which a target object is geometrically confined by robots not to escape. Caging grasps in 3D scenes are limitedly studied as several special cases because of difficulty of investigating complicated high dimensional configuration space, while 2D caging has been well formulated. We aim to examine 3D caging grasps for arbitrary shapes of object and propose an idea of discretizing entire configuration space to facilitate to explore caging sets. To discretize the configuration space of orientation, we adopt uniformly-sampled quaternions. Some simulation results verify our proposed method.