Prosthetic hand control: phase-based grasping pattern recognition using sEMG and computer vision

Wang, Shuo (2022) Prosthetic hand control: phase-based grasping pattern recognition using sEMG and computer vision. Masters thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (3MB)

Abstract

Pattern recognition using surface Electromyography (sEMG) applied on prosthesis control has attracted much attention. The strong relationship between visual perception and hand manipulation makes vision play an essential role in prosthetic hand control. Utilizing both sEMG and visual information to improve prosthetic hand control became a promising research direction. In most existing hand grasping classification research using sEMG, the signals collected during the firmly grasped period were used for classification because stable signals facilitated classification performance. However, using signals collected from the firm grasp period may cause a delay in controlling the prosthetic hand. Targeting this issue, we explored a new way for grasp classification using signals collected before the firm grasp. We examined accuracy changes during the reaching and grasping process and identified an sEMG sweet period, starts at 1100 ms and ends at 1400 ms in the early grasping phase, that can leverage the grasp classi- fication accuracy for the earlier grasp detection. Although Surface Electromyography (sEMG) achieved a feasible solution in a laboratory environment, the classification accuracy is not high enough for real-time application. Researchers proposed integrating sEMG signals with another feature not affected by amputation. The muscular coordination between vision and hand manipulation makes us consider including the visual information in prosthetic hand control. In this study, we identify another sweet period, starts at 0 ms and ends at 320 ms during the early reaching phase, in which the vision data could better classify the grasp patterns. Moreover, the visual classification results from the sweet period could be naturally integrated with sEMG data collected during the grasp phase. After the integration, the accuracy of grasp classification increased from 85.5% (only sEMG) to 90.06% (integrated). Knowledge gained from this study encourages us to further explore the methods for incorporating computer vision into myoelectric data to enhance the movement control of prosthetic hands.

Actions (login required)

View Item