Investigation on individual adaptive driver state diagnosis method based on driver model in car-following situation

Abstract

Highway fatalities in Japan have been increasing over the past 3 years despite decline of traffic fatalities. Most of accidents on highway are caused by careless driving including drowsy driving. Therefore, a number of researches on detecting such drowsy driving have been carried out, and various detection methods have been proposed: steering behavior, heart rate, etc. Especially focusing on such drowsy driving detection, this paper proposes a model-based method evaluating driving behavior using driving data which can be measured with existing on-board vehicle sensors in terms of ease of implementation. The model-based method can eliminate influence of driving environment so it can potentially evaluate small change of driver state. According to the comparison between the actual driving data and the driver model, the error between them correlates with subjective driver drowsiness. As a result, this study exhibits the feasibility of the proposed method that it can detect the drowsy driving by using the driver models.