ИСТИНА |
Войти в систему Регистрация |
|
ИПМех РАН |
||
There are many areas of potential application for brain-computer interface technology (BCI) including robot control [Kaplan et al., 2013]. The most successful BCI technology are based on detection the P300 EEG component as a reaction on target stimuli, which vary in shape and position in the stimuli matrix [Farwell and Donchin, 1988]. In this work we propose a P300 BCI robot control technology, where the stimuli differ only in color and were placed in the same position just in eyes of a robot. Methods. The main difference of our BCI from the standard P300 BCI is that instead of symbols spatially spread in the matrix, we used 3, 5 or 7 colors flashing in same place - in the eyes of a humanoid robot NAO H25, made by Aldebaran Robotics. All 5 participants featured the same set of 3, 5 or 7 colors flashed randomly in the left or right eyes separately ten times each. Participants were instructed to count the number of lighting of the selected (target) color stimulus. Each target stimuli were associated with elementary robot commands: raise right/left hand, turn right/left, etc. Results. After training session with individual classificator generation, the subjects were asked to activate thru BCI a given sequence of actions to the robot, for example, turn to the right, raise the hand, make one step forward, turn to the left, etc. In total each subject made up to 164 commands. It was shown that in average all well trained and highly motivated subjects were able to perform the robot control with the reliability 83%, 89% and 78% in the case of 3,5 and 7 colors flashing respectively. Conclusions. In this work we show that P300 BCI paradigm can work efficiently even if the stimulus differ only in color and located directly in robot eyes. This eye-to-eye P300 technique could be used to manipulate a different kind of humanoid robot.