Identification of Locomotion Modes Based on Artificial Intelligence Algorithms Using Surface Electromyography Signals

Today, Electromyography (EMG) signals have many applications in the engineering sciences, especially in biomechanics and prostheses, which normalizes the lives of people with disabilities. The purpose of this work is to evaluate the accuracy of various Machine Learning algorithms for classifying different locomotion modes using the EMG signal, from which the signal features including Mean absolute value (MAV), Root mean square(RMS), Waveform length(WL), Variance(VA) and Zero crossing(ZC) are extracted. K-Nearest Neighbor (KNN), Support Vector Machine (SVM), and Multi Lyer Perceptron (MLP) are employed as classifiers. Our features that extracted from dataset are devided into the parts: Training, Validation, and Test. The results yield the MLP algorithm with adam optimizer had the highest accuracy among other algorithms with an accuracy of 0.992. This satisfying accuracy along with the fast prediction speed of MLP algorithm make this method an appropriate candidate for online applications such as control of rehabilitation devices.