Using a SSVEP-BCI to command a robotic wheelchair

Abstract

This work presents a Brain-Computer Interface (BCI) based on the Steady-State Visual Evoked Potential (SSVEP) that can discriminate four classes once per second. A statistical test is used to extract the evoked response and a decision tree is used to discriminate the stimulus frequency. Designed according such approach, volunteers were capable to online operate a BCI with hit rates varying from 60% to 100%. Moreover, one of the volunteers could guide a robotic wheelchair through an indoor environment using such BCI. As an additional feature, such BCI incorporates a visual feedback, which is essential for improving the performance of the whole system. All of this aspects allow to use this BCI to command a robotic wheelchair efficiently.