Open eyes influence on electroencephalography signals for constructing neural network classifiers as mobile robot control brain-computer interface

Abstract

This study aims to develop an EEG-based BCI using multilayered neural networks for mobile robot control, which you can utilize with open eyes actually watching the robot's behavior. In this study, EEG measurement experiments were conducted under the following experimental conditions: subjects closed or opened their eyes while they were imagining a desired mobile robot movement. To examine influence of open eyes on EEG signals, the authors trained multilayered neural networks for EEG classification using the EEG signals measured in the experiments, and confirmed that opening eyes had a negative effect on the classification performance. Furthermore, the experimental results showed that you need to take into account the date you recorded EEG signals in the training process of neural networks.