Self-Adaption Soft Sensor Array Design for sEMG Control

Abstract

Myoelectric sensors are becoming indispensable to detect myoelectric signals and play a significant role in various applications. This article aims to propose a soft sensor array with inflatable chambers and 22 channels as a substitute for conventional disposable sensors. The new sensor is made of the soft material, which is Ecoflex00-50 and copper electrodes. sEMG signals of five gestures are collected under three kinds of sensors, which are the inflated, the uninflated, and the conventional disposable sensor, respectively. The correlation coefficients (up to 0.9), signal-noise ratio (up to 19dB), movement artifact signal, and motor unit action potential (MUAP) duration was calculated and analyzed to verify the performance of the soft sensor array. An improved method based on CNN is used to classify gestures, and the result is encouraging, which is 96.2%. The research results have supported that the inflated sensor can improve the signal quality and the classification accuracy, which can lay the foundation for the myoelectric sensor design research.