Grasp planning for everyday objects based on primitive shape representation for parallel jaw grippers

Grasping various objects is a key function required for service robots. In this paper, seven kinds of shape primitives utilized for abstracting objects to be grasped are proposed for efficient grasp planning. These shape primitives are defined in order to provide an appropriate set of grasping configurations of parallel jaw grippers before planning. Each shape primitive has basic grasping configurations, each of which is parameterized. Fist, a target object is modeled by using one or more shape primitives. Then, possible grasping configurations are pruned based on the information of task environment and robot hand to be used. After sampling of candidate grasps, the most suitable grasping position and orientation is chosen by evaluating and ranking the quality of the candidates. We show an experimental result of the application of this shape primitive based grasp planning method to a mobile manipulator.