CVT-Based Asynchronous BCI for Brain-Controlled Robot Navigation

• Published in: Cyborg and bionic systems Vol. 4; p. 0024
• Main Authors: Li, Mengfan, Wei, Ran, Zhang, Ziqi, Zhang, Pengfei, Xu, Guizhi, Liao, Wenzhe
• Format: Journal Article
• Language: English
• Published: United States AAAS 2023; American Association for the Advancement of Science (AAAS)
• Subjects: Autonomous navigation; Control systems; Human-computer interface; Intelligence; Obstacle avoidance; Roads & highways; Robot control; Robots; Segmentation; Tessellation; Voronoi graphs
• ISSN: 2692-7632; 2097-1087; 2692-7632
• DOI: 10.34133/cbsystems.0024

Brain–computer interface (BCI) is a typical direction of integration of human intelligence and robot intelligence. Shared control is an essential form of combining human and robot agents in a common task, but still faces a lack of freedom for the human agent. This paper proposes a Centroidal Voronoi Tessellation (CVT)-based road segmentation approach for brain-controlled robot navigation by means of asynchronous BCI. An electromyogram-based asynchronous mechanism is introduced into the BCI system for self-paced control. A novel CVT-based road segmentation method is provided to generate optional navigation goals in the road area for arbitrary goal selection. An event-related potential of the BCI is designed for target selection to communicate with the robot. The robot has an autonomous navigation function to reach the human selected goals. A comparison experiment in the single-step control pattern is executed to verify the effectiveness of the CVT-based asynchronous (CVT-A) BCI system. Eight subjects participated in the experiment, and they were instructed to control the robot to navigate toward a destination with obstacle avoidance tasks. The results show that the CVT-A BCI system can shorten the task duration, decrease the command times, and optimize navigation path, compared with the single-step pattern. Moreover, this shared control mechanism of the CVT-A BCI system contributes to the promotion of human and robot agent integration control in unstructured environments.