Foot Postures Classification using sEMG Signals

Abstract

The paper proposes an approach to classify three target foot postures from sEMG (surface electromyography) signals measured around right lower leg. A band-type fabric sensor is utilized to acquire sEMG signals for training and realtime testing, respectively. To implement a classifier of target foot postures, a machine learning algorithm using multi-layer perceptron (known as an artificial neural network) is utilized for the sEMG signals. Experimental result shows that the proposed scheme is effective with an overall accuracy 96%.