Affordance-based grasp planning for anthropomorphic hands from human demonstration

Abstract

Robust grasping and manipulation using multi-finger hands remains one of the key challenges of service robotics. So far, most theoretical approaches and simulators have concentrated on the search for static stable grasps, but without consideration of the task context or the different grasp types to be used during object manipulation. In this paper we describe a simple approach to derive human-like grasps for anthropomorphic hands from human teleoperation on a set of test objects. Experimental data is presented for the Shadow Dexterous Hand and eight different precision grasps that explore a large part of the reachable hand workspace. Analysis of the data shows a surprisingly simple correlation of the hand-object poses with the object size. Given a target object of known size, our algorithm provides a suitable hand-object pose and approach direction, and both the finger pre-shapes and approximate grasp-shapes required for grasp execution.