Hand gesture recognition from EMG signal using machine learning

Abstract

This work addresses the issue of upper limb gesture recognition based on the analysis of surface EMG signals recorded in the wrist area. Machine learning classification algorithms, especially LDA, are utilized for this purpose. The research focuses on future applications of this technology, particularly in the field of remote control of electronic devices in real-time, such as controlling smart home systems, robots, or other intelligent systems. In this paper, LDA machine learning classification models with different parameters were trained, achieving success rates up to 94.15% on testing data. Furthermore, a detailed analysis was conducted to examine how the specifically placed recording electrodes affected the success rate of the resulting machine learning model. This analysis can be followed to reduce the number of recording electrodes while minimizing the decrease in classification success rate.