An EMG-Based Teleoperation System with Small Hand Based on a Dual-Arm Task Model

Abstract

This paper proposes an EMG-based small prosthetic hand for teleoperated tasks. In the proposed method, a small robot hand with five fingers, which is typical of mammals including humans, is made by a 3D printer and can be controlled according to the user's intended motions based on the discrimination results of the measured EMG signals. Tasks performed by the proposed system are described by a Petri net and each state generates a revised vector of discrimination results to assist the operation by dual-arm motion discrimination considering which motions should be used for each operation. The state transitions are performed by using dynamic and static information for both forearms simultaneously. In the experiment, the participant was asked to control the proposed system and could voluntarily manipulate the targeted tasks.