Classification of Hand-Movements using MYO Armband on an Embedded Platform

Abstract

The study proposed the classification and recognition of hand gestures using electromyography (EMG) signals with aim of controlling upper-limb prosthesis. The representation is based on measuring the EMG signals through an embedded system by using a wearable band of MYO gesture control. To observe the behavior of these change movements, we acquired the EMG data of 4 healthy male subjects performing 4-upper limb movements. After extracting EMG data from MYO, we applied the supervised classification approach to recognize the different hand movements. The classification is performed with 5-fold cross validation technique under the supervision of QDA, SVM, Random forest, Gradient Boosted, Ensemble (Bagged Tree) and Ensemble (Subspace KNN) classifier. The execution of these classifiers shows the overall accuracy of 83.9% in case of Ensemble (Bagged Tree) which is higher than other classifiers. This study also gives a comparative analysis of thirteen comprehensive and most up-to-date EMG feature signals in Time-domain and Frequency-domain. To be a successful classification of these EMG features in both domains, we prefer attribute selected classifier as it gives the better performance and higher rate of accuracy i.e. 93.8%. The experimental results prove that features with (TD) are superfluity and redundant while features in case of frequency-domain show the ultimate dominance and signal characterization. The results of this study also inferred the operations which were easy for hand recognition and can be used for developing a powerful, efficient and flexible prosthetic design in future.