Telepresence control of humanoid robot via high-frequency phase-tagged SSVEP stimuli

This paper presents a high-frequency steady-state visual evoked potential-based model for a brain-controlled humanoid robot. An advantage of this model is to reduce subjects' fatigue by using visual stimuli with a frequency of 30Hz. This study optimizes the stimulus patterns to increase the brain signals and applies a fuzzy-based classification approach to identify human mental activities and convert them into control commands. Seven subjects successfully navigated a NAO humanoid robot to walk through a map with obstacle avoidance based on live video feedback. The on-line robot navigation experiment reached the average control success rate of 94.26% and an average collision of 1.8 times during a mission.