Surface Electromyographic signal based finger prosthesis control using ANN

Abstract

This paper represents the development of surface Electromyographic (sEMG) signal-based finger prosthesis control. A filter & amplifier circuit captures the EMG signal from the surface of the human hand that can be recorded using ATmega-2560 micro-controller. The analysis of the output signal is done to study time domain features. In this paper, standard deviation, mean, a variance is taken as time domain feature. The signal is then trained using simple Artificial Neural Network to classify accurately two finger motion i.e. grip motion and thumb index finger motion.