Research on Recognition of Forearm sEMG Signal Based on Different Motion Modes

• Published in: 2018 IEEE International Conference on Cyborg and Bionic Systems (CBS) pp. 581 - 584
• Main Authors: Fu, Menglong, Xue, Jinwei, Huang, Pingao, Chen, Zhenxin, Wei, Wenhao, Li, Guanglin, Chen, Shixiong
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.10.2018
• Subjects: Data acquisition; Electrodes; Electromyography; Muscles; Pattern recognition; Robots; Wrist
• DOI: 10.1109/CBS.2018.8612195

Pattern recognition of limb motion based electromyography (EMG) is a core technique in active prosthesis control. By decoding different EMG signal, we can obtain different motion intentions. As a kind of EMG signal, surface electromyogram signal (sEMG) signal are widely used in research for its non-invasive characteristic. How to collect sEMG signal for recognition effectively is momentous part for EMG signal decoding. For this, we design two methods of collecting sEMG signal, which according to the situation in the experiment. One method is to repeatedly collect the same action until all the interested actions are collected. Another method of sample is to acquire the interested movements randomly. Ultimately, we find that the average recognition rates of these two methods are 96.9% and 97.5% respectively. These two methods have no significant effect on the final classification results. Therefore, in the experiment, we can take a more convenient method of continuous action acquisition to obtain sEMG signal. At the same time, the classification results are sent to the control system for verification, which we obtained a high quality recognition results. This also indirectly verifies the usability of the EMG acquisition system we designed.