Study On Rear-end Collision Warning Algorithm Based On Driver Characteristics

Automobile brings us both great conveniences and safety problems couldn’t beignored. Road traffic accidents have become the first killer leading to human’s irregulardeath, and the rear-end accidents typically accounts for more than30%of all trafficaccidents. Therefore, a variety of rear-end collision warning algorithm are developed byresearch institutions at home and abroad in recent years. But most existing algorithms arebased on safety distance model. In another word, they provide alarm to drivers based onthe distance between vehicles acquired by radar or visual sensor. However, in this process,the importance of man-machine interaction is always neglected. Researches show that it isnecessary to ensure the algorithm’s function matches to the driver’s characteristic toimprove driving safety without interrupting driver’s normal performance. Hence, it issignificant to set up a rear-end collision warning algorithm meeting driver’s characteristicsfor reducing rear-end collision accident.By analyzing and summarizing the existing car rear-end collision warning algorithmof domestic and foreign, the driver braking behavior and reaction time is studied, anddrivers are divided into9types based on the two aspects.5parameters could representdrivers’ car-following expectation are selected as the input, and3different kinds ofcar-following state as the output. Then different collision warning discrimination functionscould be established for different types of driers. The established algorithm is verified inthe end. Concrete research content is as follows.1. Theoretical analysis and test scheme design. Firstly, the driver braking behavior andreaction time is studied, and the significant influence of them to the car safety is clarified.Secondly, on the basis of the analyzing the research status at home and abroad, test schemeis designed combining with the research purposes and demands, including the installationand debugging of the test facility, the development and application of the scenario, therecruitment of the testers and the selection and processing of the parameters. All the abovecould provide a theoretical basis and data support for the following research.2. The classification of drivers. First of all, the moment, when the acting forces on theaccelerator or brake pedal changes fastest, is determined as the first reaction time after thefront vehicle’s emergency braking through theoretical analysis. The time differencebetween them is the driver’s reaction time. Based on the reaction time, K-averageclustering algorithm is utilized to divided drivers into sensitive type and normal type andinsensitive type. Then, with selecting several parameters which could represent the driver’sbrake control behavior, drivers are divided into radical type, normal type and conservative type on basis of fuzzy C-means clustering algorithm. Finally, by considering the above twocategories, all drivers could be divided into nine categories.3. Establishment of the algorithm. In the first place, after each type of driver’s controlparameters sorting separately,5parameters, which could represent driver’s car-followingexpectation, are achieved with selection and calculation. Then, the front car emergencybraking moment is regarded as the start time of dangerous state. Based on Fisher criterion,the5above parameters are selected as the input, and safety, dangerous and very dangerousthree car following state as the output, eventually the rear-ended warning discriminationfunction is achieved. With the same method discrimination functions fitting for each kindof drivers are obtained. At last, the established algorithm is verified.With consideration of driver’s characteristics in the rear-end collision warningalgorithm, the established algorithm matches driver’s behavior and expectation, andacquires better accuracy, effectiveness and acceptability. The research achievement couldprovide theoretical basis and technical support for the study and application of rear-endcollision warning system, then improve road safety and reduce traffic accidents.