Classification of gait direction using brain signals

The use of brain-computer interface (BCI) technology has emerged as a promising rehabilitation approach for patients with motor function and motor-related disorders. Brain recordings collected by an electroencephalography (EEG) system have been widely employed in BCI platforms including assistive robots such as exoskeletons. Despite the large body of literature on upper-limb exoskeleton, few studies were dedicated to lower-limb exoskeletons. The current literature is limited to decoding the EEG rhythms to identify gait intention and gait variations to ultimately command a lower-limb robotic exoskeleton. This experimental study, however, aims to decode the EEG signals for the actions of gait toward four different directions and stop. Three different deep neural network paradigms were used to generate distinct classification models. Ten-fold cross-validation scheme was employed to compare the classification performance obtained for each of these models. The outcome of this study has the potential to be ultimately used in lower-limb exoskeletons for interactive navigation in robotic rehabilitation therapy.