A multifunctional prosthesis control system based on time series identification of EMG signals using microprocessors

Abstract

This paper describes a real-time system for separation among several limb functions, in order to provide multifunctional control of an upper-limb prosthesis for above-elbow amputees. The system employs microprocessor hardware and is based on identification of voluntary myoelectric signals resulting from isometric contractions of the musculature of the residual limb, and on subsequent discrimination of these signals for control of the several degrees of motion of the prosthesis. The system requires only one to two electrode sites. (Contrary to the usual placement of electrode pairs directly over specific muscle bellies, to eliminate crosstalk, we prefer to place our electrode pair between muscles so as to acquire the different weakly-correlated signals associated with each of several different voluntary functions). The system satisfies the various practical constraints of weight, volume, and speed, as arise in practical prostheses. Preliminary amputee tests on the system have resulted in an 85 percent success rate using 8-bit double-precision micro-computer hardware.