MULTIMODAL APPROACH BASED ON PHYSIOLOGICAL SIGNALS FOR DETECTION AND RECOGNITION OF THE LOWER LIMB MOVEMENTS

Employing biosignals to capture the user's intentions behind the motion via a human-robot interface (HRI) is a promising technique in the domain of rehabilitation and assistive robotics. Some of these HRI systems are based on brain signals such as electroencephalogram (EEG) or functional near-infrared spectroscopy (fNIRS) whereas others rely on a myoelectric signal such as an electromyogram (EMG). EEG signal is one of the most common biosignal used in the rehabilitation and assistive robotics realm. However, EEG suffers from some issues such as low detection accuracy and low spatial resolution of the EEG signal that results in a redundant channel. Therefore, there is a need to integrate the EEG signals with other biosignals such as fNIRS and EMG signals to increase the detection accuracy and select the most related channels to the movement task. This research aims to develop a multimodal approach based on the fusion of the biosignals to detect and recognize lower limb movements. The first aim is to select the optimal EEG channels that are related to the lower limb movements.