Implementation of brain-computer interface based on SSVEP for control of a lower-limb exoskeleton

Abstract

In this paper, we present a control method that combines Brain-computer interface based on Steady-State Visual Evoked Potentials and traditional force/angle sensors to control a lower-limb rehabilitation exoskeleton. Through visual stimulation experiments, the subjects will produce raw electroencephalogram signals including commands information. After acquiring such information, the control system of exoskeleton robot first pre-process it by using butter-worth filter and Fast Fourier transform. By means of canonical correlation analysis, the system then matches the signals with the template signal, in order to decode the electroencephalogram signals and obtain subjects' intentions. At the same time, the robot will judge its state through its own sensor system. Finally, the control system executes the corresponding action after combining status and commands information. The command recognition accuracy is about 91% during the offline experiments. Adding states of the robot could also reduce the probability of robot malfunction during the online experiments. The results verified that the proposed method combining Brain-computer interface and traditional force angle sensor could be effective to control lower-limb exoskeleton robots.