A Lower-Limb Exoskeleton Control System based on Brain-Computer Interface with Multiple Motor Imagery Segments Decoding

With the persistent exploration of brain science, brain-computer interfaces (BCI) have been developed and applied in various fields in recent years. Lower-limb rehabilitation exoskeleton control systems based on BCI achieve better effects than traditional rehabilitation methods. However, control systems based on motor imagery (MI) signals are still difficult to popularize because of low online recognition rate, high latency and unstable control. In this paper, a lower-limb exoskeleton control system based on multiple segments decoding is proposed, which can support the subject to complete the movement of raising unilateral leg tasks. First, the MI decoder is trained by the data collected offline. The online control system wirelessly connects the EEG cap with the lower-limb exoskeleton, transmits the MI signal in real-time for online decoding, and controls the exoskeleton according to the decoded instructions, which a multi-segment decoding strategy is used to improve control accuracy and system robustness. The effectiveness of the proposed system is evaluated in the online experiment, which indicated that the system is efficient for walking rehabilitation training under various scenarios.

I. Introduction

With the unremitting exploration of human brains, J. J. Vidal first proposed the concept of brain-computer interface (BCI) in 1973 [1]. BCI realizes the information exchange between the brain and external device by creating a direct connection between them, including invasive BCI, semi-invasive BCI, and non-invasive BCI [2]. To analyze the electrical activities of the brain, electroencephalogram (EEG) is applied widely in non-invasive BCI systems. EEG signals are generated through the cortical nerve cell inhibitory and excitatory postsynaptic potentials, which can be measured by placing electrodes on the surface of the scalp [3]. Due to its safety, convenience, and cost-effectiveness, EEG-based BCI has gradually been applied in various fields, such as neurological disease diagnosis, emotion detection, user experience evaluation, and rehabilitation [4] [5].