An intelligent prosthetic system for EMG pattern recognition based prosthesis control*

Abstract

Fitting an appropriate artificial limb is an effective method to help the Upper limb amputees to restore their motor function. In decades, scientists have conducted a series of studies to decode the motor intent of amputees to achieve naturally and dexterous control of prosthetic limbs. Among them, Movement intention by using EMG-based pattern recognition (PR) algorithms from the residual muscle electrophysiological information of the limbs is one of the important methods of multi-functional prosthetic control. However, the EMG-PR based prosthesis is still not widely applied in clinical. One of the major reasons is the real-time performance of various PR algorithms is not robustness enough for clinical application since laboratory performance metrics used to evaluate PR algorithms may be poorly associated with the clinical outcomes. And another important reason is the lack of bio-imitability of prostheses, because the developed bionic prostheses with five independently controlled fingers are heavy and lack of active wrist. Therefore, this paper proposes a multi-degree-of-freedom intelligent prosthetic system for EMG PR algorithms assessment, including the PR algorithm system software named REHPS, a Real-time Control System for Prosthesis (RCSP) and a lightweight prosthetic hand named S-HAND. Both able-bodied subjects and amputees could wear the prosthetic system to carry out tests which are mimicking the daily use of prosthesis. These tests would provide clinic performance metrics for the research of motion intent recognition algorithm. The system integrates advanced motion intent recognition algorithms into the prosthesis and also provides more effective evaluation methods for myoelectric pattern recognition algorithm research.