Abstract: Fatigue is one of the important causes of traffic crashes, and effective fatigue detection models can reduce traffic collisions. There are large individual differences on fatigue performance and significant cumulative effect on driving fatigue over time. However, previous studies were insufficiently evaluated individual differences and time cumulative effect of fatigue driving. The purpose of this study is to develop a fatigue detection model that considers individual differences and time cumulative effect of fatigue. Fifteen drivers’ driving behavior and eye movement data were collected by a driving simulator with an eye-tracking system. Drivers’ subjective fatigue score was collected by the Karolinska Sleepiness Scale. As a times series model, the state space model (SSM) considers the time cumulative effect of fatigue. Moreover, SSM does not require large dataset to have high accuracy, besides handling missing data. Therefore, this study used SSM as the fatigue detection model. Under this guidance, the SSM was built for each driver to retain individual features. The difference in results between each SSM show that there is a performance diversity among drivers. Being able to take personalized information from every single driver into account makes individual modeling more suitable for fatigue detection compared to other models that integrates multiple driver data. Results proved those points through detecting fatigue with a detection accuracy of 75.94%. To sum up, the advantages mentioned above make SSM a comprehensive and valuable individualized fatigue detection model for commercial use.
Xuesong Wang*, Mengjiao Wu, Bowen Cai, Chuan Xu, Xuxin Zhang, Xiaohan Yang. A State Space Detection Model Considering Individual Differences and Time Cumulative Effect of Driving Fatigue.Transportation Research Board 101st Annual Meeting, Washington D.C., USA, 2022. 1.9-13.