Research on Estimation of Driving Behavior Based on EEG Signals

Alertness is an ability to measure peoples attention in a certain time. And the alertness can be estimated by observing the response time and sensitivity of person who doing tasks for a long time. In a lot of research, a lack of alertness has been implicated as a major factor in road accidents. We first build a simulated driving platform using Unity3D and collect 10 EEG signals from frontal and occipital regions and information of simulated driving offset. Then, we get moving-averaged power spectral with EEG signals by the sliding window technique, calculate the correlation with the driving offset, and find the highest correlation of channels is FP1POZ and OZ. Finally, we extract features by principal component analysis, construct regression models, and predict simulated driving offset. The result shows that the behavioral data can be simulated by EEG signals and the alertness status can be tested by EEG signals, when the drowsy state of subjects appears more concentrated.