Towards Early Status Warning for Driver's Fatigue Based on Cognitive Behavior Models

摘要

Based on ACT-R (Adaptive Control of Thought-Rational) cognitive architecture this paper implements researches on a status warning system for driver's fatigue, its goal is applying vehicle performance output and cognitive science to build driver behavior model, using non-invasive detection method that retrospect driver behavior based on model to monitor driving status, and to reach the aims of driver status monitor and early warning. First, based on the different detection methods' analysis of driving fatigue, the predominance of cognitive science, and the inherent relationship between driver behavior and cognitive science, the advantages of applying cognitive theory to researches on driver fatigue are clarified. Then, based on the analysis of the factors contribute to fatigue related accidents and observations of drowsy driving cases, the viewpoint that the fatigue driving is consist of three stage and corresponding to three status is proposed; accordingly, the declarative and procedure knowledge for ACT-R architecture is extracted, and driver fatigue behavioral model is implemented on the ACT-R software platform. Finally, the simulation methods are applied to verify the model's validity and a framework of driver status monitor and early warning system that contains the cognitive fatigue driver behavior models is put forward. The research results indicate that the fatigue driver behavior model has a strong advantage in the researches of driver status monitor and early warning.