EMG SIGNALS BASED GESTURE RECOGNITION FOR COMPUTER INTERFACE

Abstract

This paper presents an electromyogram (EMG) signals based hand gesture recognition for computer interface using an inside of the forearms-placed dual channel EMG sensor (shimmer 3). EMG signals were gathered from 10 participating subjects after filtration and normalization and to improve the data set used to train the machine -learning model, the system uses Time-Domain (TD) features and Frequency-Domain (FD) features of the signals that were collected. To develop a classification model of the signals produced by four different gestures performed with two hands, classification learner app in MATLAB was used along with an Ensemble Boosted Trees learning model with the chosen features, the Bayesian optimizer was then used and result with an accuracy of 84.1%. An RC car was used as a sample application so that it could be operated by the user's hands EMG signals rather than a remote control device.