Driver Distraction Detection Using Bidirectional Long Short-Term Network Based on Multiscale Entropy of EEG

Driver distraction diverting drivers’ attention to unrelated tasks and decreasing the ability control vehicles, has aroused widespread concern about driving safety. Previous studies have found that performance decreases after used vehicle behavioral features detect distraction. But how brain activity changes while remains unknown. Electroencephalography (EEG), a reliable indicator of activities been widely employed in many fields. However, challenges still exist mining information EEG realistic scenarios with uncertain information. In this paper, we propose novel framework based on Multi-scale entropy (MSE) sliding window Bidirectional Long Short-term Memory Network (BiLSTM) explore driver multi-modality signals real traffic. Firstly, MSE is implemented extract determine position. Statistical analysis data then performed validate indeed around Finally, use BiLSTM other traditional features. Our results show notably Consistent result MSE, significantly deviates from normal state Besides, outperforms entropy-based methods better than Additionally, accuracy improved again adding feature 3% increasement. The proposed useful for detection applications scenarios.