Investigation of Electrode Location to Improve the Accuracy of Wearable Hand Exoskeleton Trainer Based on Electromyography

Abstract

EMG signal has a random and stochastic characteristics, so it is difficult to predict the amplitude. Furthermore, the EMG signal depends on the electrodes location. Therefore, a proper muscle selection determines the system's accuracy value. The purpose of this study was to investigate the exact location of the electrodes to improve the accuracy of the wearable hand exoskeleton trainer based on electromyography (EMG) signal control. The main advantage of the results of this study is that the most dominant muscle was found in the development of a wearable hand exoskeleton based on an EMG signal threshold. Therefore, the model can be controlled using a single electrode pair which can further be applied using a low-cost microcontroller. In this study, ten respondents were involved in the data acquisition. The discovery of the dominant muscle was carried out by investigating the dominant EMG signal in three muscles (Abductor pollicis longus, extensor digitorum) that plays a role in the open and close movements of the hand exoskeleton. Dry electrode was used to detect EMG signal activity on the skin surface. The EMG signal was then extracted using the root mean square (RMS) feature. After the evaluation, the results showed that the flexor digitorum superficialis muscle in the rest position produced higher accuracy value than the other muscles, which was 96.63±0.67%. In the implementation, the product of this research can be applied for rehabilitation steps in post-stroke patients which is carried out either in a medical rehabilitation unit or at home independently.