Research On Driver Visual Distraction Detection Technology

Driver distraction has been a major concern of the traffic accident, since it reducing driver’s information acquisitioning ability and prolonging reaction time of emergency accidents. As use of mobile communication devices and in- vehicle information systems(IVISs) such as cell phones, navigation systems, and satellite radios has increased, visual distraction is one of the common problems which threaten driving safety and people’s life. Hence, developing a driver visual distraction detection technology in urban traffic environment is significant to improve driving safety and traffic environment.First of all, depending on different types of distraction, this paper analyzed the status quo of driver distraction detection methods both in China and abroad, established the database based on a driving simulator experiment. Suitable area of fixation division approach for visual distraction detection was taken based on driver’s visual characteristics and driver’s behavior. Then, after comparing and statistical analyzing the visual and vehicle dynamics data from normal driving and visual distraction driving, several key characteristic parameters were extracted. After that, driver visual distraction detection model were developed based on a machine learning method-Support Vector Machines, further more, another multiple characteristic method considering long period fixation was proposed. Finally, the two models were evaluated for the detection results.