Control an exoskeleton for forearm rotation using FMG

Abstract

In the field of robotic rehabilitation, surface electromyography (sEMG) has been proposed for controlling exoskeleton device for assisting different movements of the human joints, such as the shoulder, the elbow, the wrist and the fingers. However, few works have been proposed for using sEMG to control a forearm exoskeleton for assisting the movement of pronation and supination. The main difficulty for employing the sEMG control approach is the low signal to noise ratio of the pronator and supinator muscle group. To overcome this difficulty, we propose an alternative method utilizing force myography (FMG) instead of the sEMG for controlling a forearm pronation/supination exoskeleton. An easy setup strap with an array of force sensors was developed to capture the forearm FMG signal. The FMG signal was processed and classified using the state-of-art machine learning algorithm - Extreme Learning Machine (ELM) to predict the forearm position. The prediction results can be used to control a forearm pronation/supination exoskeleton. A bilateral experiment with two protocols was designed to demonstrate one of the potential applications of the proposed system, as well as to evaluate the system performance in terms of classification accuracy. One volunteer participated in the experiment. The result shows the system was able to predict the position of the forearm using the proposed method with 98.36% and 96.19% of accuracy.