The design of simulation vehicle system controlled by multichannel EEG based on imaginary movements

This paper focus on the simulation experiments of controlling a car by EEG based on imaginary movements. We collected fourteen channels of brain electroencephalogram (EEG) signals transmitted by Emotiv Epoc+ Neuro Headset. Instead of emotiv's own processing and analysis methods, in our program, we wrote wCCA algorithm to preprocess EEG signals, and used Common Spatial Pattern (CSP) to extract the feature of EEG. Then, by KNN algorithm we classified the extracted features as imagining feet & righthand movement and neutral state, which correspond to the three control commands (turn forward, turn right, stop) respectively, so that it can control the motions of the simulated car constructed in SigViewer2.2.0. In addition, we applied the method of online-learning to greatly improving the classification accuracy, which avoid the time-consuming and laborious off-line training and is closer to real-time status of subjects. The simulation experiments results show that the methods of processing and classification in EEG have better feasibility and stability. It establishes a basis of practical application for EEG-controlled automobile.