Cyber-Physical System Control Based on Brain-Computer Interface

The study describes approaches of direct and supervisory control of cyber-physical systems based on a brain-computer interface. The interface is the main component of the control system, performing electroencephalographic signal decoding, which includes several steps: filtering, artefact detection, feature extraction, and classification. In this study, a classifier based on deep neural networks was developed and applied. Description of the classifiers based on convolutional neural network is given. The developed classifier demonstrated accuracy 73 ± 5% of decoding four classes of imaginary movements. Prospects of using non-invasive brain-computer interface for control of cyber-physical systems, in particular, mobile robots for maintenance of immobilized patients and devices for rehabilitation of post-stroke patients are discussed.