Improving Reliability of Life Applications Using Model-Based Brain Switches via SSVEP.

Abstract

The brain switch improves the reliability of asynchronous brain-computer interface (aBCI) systems by switching the control state of the BCI system. Traditional brain switch research focuses on extracting advanced electroencephalography (EEG) features. However, a low signal-to-noise ratio (SNR) of EEG signals resulted in limited feature information and low performance of brain switches. Here, we design a virtual physical system to build the brain switch, allowing users to trigger the system through periodic brainwave modulation, fully integrating limited feature information and improving reliability. Furthermore, we designed multiple experiments to validate the effectiveness of the proposed brain switch based on steady-state visual evoked potentials (SSVEP). The results verified the performance of SSVEP brain switches based on virtual physical systems, improving the reliability of brain switches to 0.1 FP/h or even better with acceptable triggering time and calibration-free for most subjects. This represents that the proposed virtual physical model-based brain switch can utilize SSVEP features and output the reliable commands required to control external devices, promoting BCI real applications.