ドライバの運転動作に基づく個人特性を考慮した危険運転予測システムの構築

Many car accidents are caused by driver’s deviation from normal condition like carelessness. We aim to construct a driving assist system that is able to detect driver’s deviation signal from normal condition. The system detects the deviation signal using driver’s time-series head motion information. In this paper, we optimize categorization of driver’s head motion using two kinds of unsupervised neural networks: Self-Organizing Maps (SOMs) and Fuzzy Adaptive Resonance Theory. Moreover, we introduce a method to analyze an individual difference in an electroencephalogram (EEG) using a SOM. In order to confirm the individual difference in the EEG, we conduct experiments using real EEG data. The experimental results suggest that it is possible to express the individual difference using SOM.