Control Strategy for the Development of Bio-Orthotic Limbs Using EMG Signals

This paper elucidates a control mechanism for bio-orthotic limbs, analyzing electromyogram (EMG) signals, to improve the response time and efficiency. The dataset consisting of ten different classes of finger movements is used in the study. Various features are extracted from the trials to obtain temporal as well as frequency domain information. To increase the response time of the model, different feature reduction techniques are discussed. The reduced feature set is used to explore two different classification methods. The highest classification accuracy of 97.06% was obtained using Random Forest Classifier when compared with 76.49% outcome of support vector machine classifier.