Status Identification and Object In-Hand Reorientation Using Force/Torque Sensors

Abstract

Object in-hand reorientation is widely known to be crucial for robotic dexterous manipulation, and the stable reorientation of the in-hand object with the two-fingered gripper is challenging. In this paper, the object reorientation is completed using a commercial two-fingered gripper with integrated force/torque sensors. The identification models for the object in-hand status (slipping status and pivoting status) are built based on the friction theory. In-hand status identification provides an important guarantee for the object in-hand reorientation. Moreover, we propose a control framework for the object in-hand reorientation under the action of gravity. This framework is to make the object slip or pivot by continuously controlling the gripping force. The force control law is related to the object's kinematics parameters, and the gripping force is modified according to the object in-hand status and kinematics parameters. Finally, the effectiveness of the proposed method is validated by testing object slipping and object pivoting in experiments. Results show that the gripping point and pose of the in-hand object are adjusted through object slipping and object pivoting.