Driver State Identification On The Freeway And Research On Automobile Anti-collision Warning System

With the development of the fast traffic network, people’s travel is getting more alternative and convenient, but the problem on traffic safety is getting more serious. According to the investigation and analysis made by the expert from American Expressway Administration, the accidents in traffic accidents due to personal factors account for 90%, the traffic accidents were caused by vehicle factors account for 3%. In China, over 50% of the expressway traffic accidents were caused between vehicles, among them, the accidents due to automobile rear collision account for 35%, the main causes of rear collision are the driver negligence. Especially in the expressway, it is easy for driver negligence to bring frightful calamity. Therefore, the discrimination for the driver behaviors and the driver state is valid to improve collision warning technology and induce traffic accidents. This paper will have a study on the real-time identification of driver state and the modeling longitudinal safe distance. Details are as follows:Firstly, according to the existed research achievement analysis the relationship between the driver’s state with physiological signals, driving behaviors and eye features, determine the critical value of the heart beat rate signals, steering signals and eye features, which can provide a theoretical basis to identify the driver’s state.Secondly, research of the driver’s eye detection algorithm. The eye detection is directly concerned with the identification of driver state. This paper presents a new algorithm which is fusion of a mixed color model, integral projection and Canny algorithm, establishes a mixed color model to locate faces in color space, and finds face area through integral projection and formwork design. Then eye area is detected through integral projection and a curve optimal. In the end, localizes eyes accurately using Canny algorithm and morphologic processing. The accuracy is more than 90%.Thirdly, driver’s state identification on the freeway, fatigue and distraction are easy for drivers on the freeway. A driver state identification model was established based on Information Fusion. Through detecting eye feature, heart beat rate signals and steering signals, and establish a driver state recognition model by adopting information fusion technique combining fuzzy assessment with D-S evidence theory and the effectiveness of the model is verified by an example.Fourthly, establish a new safe distance model based on driver state. Deduce the kinematic safe distance model and improve it by analyzing the automotive braking process and the preceding vehicle driving state. Then through driving simulation experiments to obtain the date of different drivers’ reaction time in different states and give the parameter values of the kinematics safe distance model based on the experimental data. Finally, compare the new model to the typical safe distance model.