Robot grasp planning based on demonstrated grasp strategies

Abstract

This paper presents a novel robot grasping planning approach that extracts grasp strategies (grasp type, and thumb placement and direction) from human demonstration and integrates them into the grasp planning procedure to generate a feasible grasp concerning the target object geometry and manipulation task. Our study results show that the grasp strategies of grasp type and thumb placement not only represent important human grasp intentions, but also provide meaningful constraints on hand posture and wrist position, which highly reduce both the feasible workspace of a robotic hand and the search space of the grasp planning. This approach has been thoroughly evaluated both in simulation and with a real robotic system for multiple daily living representative objects. We have also demonstrated the robustness of our approach in the experiment with different levels of perception uncertainties.