A Novel Semantically Congruent Audiovisual Interface for Assisting Brain-Machine Interface (BMI) Performance Enhancement

Abstract

Brain-Machine Interfaces utilize distinct brain patterns as control commands. However, many BMIs suffer from low performance issue even with the state-of-the-art classification algorithms in hand. Herein, we propose a novel BMI interface using semantically congruent audiovisual stimuli involving contextual motions to assist BMI performance enhancement. We designed two motion classes of “up” and “down” using two paradigms: Visual only and congruent audiovisual pair. We first compared the level of spectral discernibility between the two given commands within each paradigm using frontal, temporal, and occipital channels. We then applied these paradigms onto a EEG-controlled drone system. Although the power spectral density did not show any statistically significant differences, the subjects’ drone controlling performance increased by 16% with the audiovisual interface compared to the visual only interface. Thus, this semantically congruent audiovisual BMI interface using contextual motion stimuli may be used as a supportive tool for enhancing BMI performance.