Dexterous Hand Development with Full-Hand Tactile Perception and Learning-based Grasp Stability Assessment

Abstract

In this work, the idea of full-hand tactile perception is proposed. A 4-fingered fully-actuated dexterous hand with 16 degrees of freedom (DoFs) is designed, and a novel and facile piezoresistive flexible tactile sensor is proposed and integrated on the index finger, thumb, and palm. Moreover, a graph convolutional network (GCN)-based grasp stability assessment system is introduced as a preliminary usage of full-hand tactile perception, where envelope grasp is taken into consideration along with traditional 2-finger grasp. Experimental results show precise hand movement and perception, with grasping stability evaluated correctly.

Mingqi Chen

Postdoctoral Researcher

Zhenyuan Zhang

Dexterous hand design

Shukun Huang

Student

Qiang Li

Professor and header of AG

My research interests include underwater robot, collaborative robots, humanoid robots.