Classification of EMG signals through wavelet analysis and neural networks for controlling an active hand prosthesis

Abstract

In order to increase the effectiveness of active hand prostheses we intend to exploit electromyographic (EMG) signals more than in the usual application for controlling one degree of freedom (gripper open or closed). Among all the numerous muscles that move the fingers, we chose only the ones in the forearm, to have a simple way to position only two electrodes. We analyze the EMG signals coming from two different subjects using a novel integration of ANN and wavelet. We show how to discriminate between more movements, five in this study, using our new classifier. Results show how the methodology we adopted allows us to obtain good accuracy in classifying the hand postures, and opens the way to more functional hand prostheses.