Design and Implementation of a Decision Making System for Controlling a Hand Exoskeleton Based on EEG/EMG Signals

This paper presents an approach of combining Electroencephalography (EEG) and Electromyography (EMG) signals to create a hybrid Brain Interface Computer (BCI) device for controlling a hand exoskeleton through classifying the flexion attempt and resting states of the subjects. We analyzed data of 51 healthy and patients with brain lesions that involved the motor cortex and different hand movement disorders. Their EMG and EEG activity were recorded while the subjects attempted to move their index finger. The signals are analyzed through deep neural network, restricted Boltzmann machine and LDA classifier methods to access a reliable accuracy. Further, to evaluate the proposed approach, we designed and implemented a robotic system for rehabilitation of the hand movement to show that it is able to derive assistive control, by detecting flexion movement from the signals as a command and send it to robot. The system is combined with control signals that determine the information of flexion according to measured EMG signals.