EMG-controlled hand exoskeleton for assisted bilateral rehabilitation

Abstract

This article presents an electromyography (EMG) controlled hand exoskeleton for basic movements in assisted bilateral therapy, where bimanual work is required by the user. The target users are individuals with the right hand affected by an accident or cerebrovascular problems which require passive or assisted rehabilitation. Through a Matlab GUI, the system receives, processes and classifies electromyographic signals from the user acquired by a MYO armband obtaining an accuracy of 81.2% using k-Nearest Neighbors (kNN) as the classification algorithm and Random Subset Feature Selection (RSFS) as the feature selection algorithm. Subsequently, the exoskeleton reproduces the movement detected in the user’s opposite hand. The exoskeleton prototype is 8 degrees of freedom (DOF), built using 3D printing and has independent movement of the fingers. The movement controller is based on fuzzy logic. For the system performance analysis, kinematic information from a motion capture system is used to compare the trajectories in different grasping tasks of a user’s hand with and without the exoskeleton with a maximum error of 10.63% and a minimum of 3.46% with the desired final position, which physically represents a difference of 1.89° and 0.07° respectively.