Applicability of SSVEP-based brain-computer interfaces for robot navigation in real environments

Abstract

Brain-computer interfaces have been extensively studied and used in order to aid patients suffering from neuromuscular diseases to communicate and control the surrounding environment. Steady-state visual evoked potentials (SSVEP) constitute a very popular BCI stimulation protocol, due to their efficiency and quick response time. In this study, we developed a SSVEP-based BCI along with a low-cost custom radio-controlled robot-car providing live video feedback from a wireless camera mounted on the robot, serving as our testbed. Our goal was to quantitatively assess the applicability of SSVEPs in real time navigation in realistic environments using a pragmatic approach. In order to assess the additional fatigue that the camera video introduces, we designed a two-session experiment, a control one with no connection to the robot and, thus, no live camera feed, and a realistic one where the users could navigate the robot with the provision of front scenes, captured from the camera. Statistical tests revealed a significant decrease of the accuracy of the system during the realistic session that included live video, in comparison with the session that did not. The results suggest that the moving camera image sequence introduces an extra level of fatigue and/or distraction to the users.