Towards better real-time control of smart wheelchair using subtle finger movements via wireless (blue-tooth) interface

This study investigates the design of a novel realtime system to control a smart wheelchair using subtle finger movements. Surface electromyography (sEMG) signals from the forearm recorded during muscle onset activation were used as the control signals. A set of time domain (TD) and wavelet features were extracted from sEMG signals. Wavelet singularities based on the volume conduction muscle model were computed to provide precise control to the system. A simple multilayer perceptron artificial neural network (ANN) is applied for classification of these features. The output of the classifier was used as a control signal to test a small miniaturized wheelchair in realtime via wireless (blue-tooth) interface. Wavelet decomposition singularities in association with an ANN classifier can successfully differentiate between five different subtle finger movements with a high degree of sensitivity. This research study is a framework towards providing a simple and better control and interface for amputees, disabled and elderly, who have limited mobility.