On the Brain Computer Robot Interface (BCRI) to Control Robots

摘要

Brain Computer Interfaces (BCI) have emerged as a solid, though promising yet incipient, mechanism to communicate simple and abstract high-level electroencephalography (EEG) commands by decoding brain rhythms. When BCI are used for online commands of a kinematic or dynamic external agent, either real or virtual ones, human perceives the task with different attributes, thus different brain regions are activated depending on the characteristic of the perceptual task and the cognitive process involved. Hence, there arises the question on how the perceived kinematic or dynamic task attributes affects the ability of human to produce a BCI command. In this paper, a Brain Computer Robot Interface (BCRI) is proposed, which contains as a core element a BCI to decode commands for a dynamical external agent, such as a robot. To convey the attributes of a dynamic perceptual task, advanced tools of dynamics and control of robots are borrowed to integrate the BCRI, complying with real time constraints and behaviors. Experimental results shows that the ability to generate BCI commands is due to a lower user frustration, which is obtained when the perceptual task matches the dynamic attributes exhibited by a dynamic robot. In contrast, when the robot behaves with kinematic attributes only, high frustration impedes the generation of BCI commands.