A robot prosthetic finger system based on finger joint angle estimation using EMG signals

Abstract

The goal of this study was to develop a robot prosthetic hand system for an amputee that estimates his desired finger angle from neural signals, and operates with the motion he intends. Consequently, we considered a method for estimating finger joint angle from biosignals. We have previously proposed a finger joint angle estimation method based on surface electromyography (EMG) signals and a linear model. This method was based on a histogram of EMG signals, and may be applicable to surface EMG signals as well as neural signals. To confirm this, we carried out finger joint angle estimation experiments using both surface EMG signals and needle EMG signals measured at deep layers of muscles and are similar to neural signals. Moreover, an artificial finger robot has been designed and developed for finger amputation patients. This finger robot has 2-DOF and each joint is worked with motors and cables. The joint angles are controlled by a PID method. Using this robot and applying our proposed finger joint angle estimation method, we carried out experiments in which the robot is synchronised with the subject’s third finger motion. As a result, it was confirmed that our system is technically feasible for application to a real prosthetic hand system.