Improving Myoelectric Prosthetic Arm Proportional Control under MUAP Envelope Colored Noise

Abstract

Accuracy and stability of the myoelectric prosthetic arm digital proportional control strongly depends on shape of the rectified motor unit action potential (MUAP) envelope. To mitigate effects caused by the envelope-induced torque ripples, mechanical vibrations (TRMV), and electromechanical delays (EMD), in this article we develop in discrete-time state-space the unbiased finite impulse response (UFIR) filter, Kalman filter (KF), and game theory recursive H∞ filter. The filters developed are called the cUFIR filter, cKF, and c∞ filter and applied to extract a smooth near Gaussian MUAP envelope by viewing ripples as colored measurement noise (CMN). Additional UFIR and Rauch-Tung-Striebel (RTS) smoothers are also used to improve the performance when the EMD is acceptable. Based on experimental examples of prosthetic arm control, it is shown that the advanced cUFIR filter, cKF, and c∞ filter improve the performance of the original filters by the factor of about 10 and essentially reduce the TRMV and EMD.