The Study Of Driving Tendency Identification And Modeling Longitudinal Safe Distance

Traffic accidents threaten the people’s lives and properties seriously. Statistics show that the accidents caused by the drivers were more than 90% directly or indirectly. Automotive collision avoidance-warning system (CAS) can assist the driver timely and reduce traffic accident effectively. The collision avoidance-warning algorithm as the basis of CAS decides when the system involves into. If the system involved too early, it will interfere the driver’s normal driving; if the system involved too late, collisions cannot be avoid completely. However, the existed collision avoidance-warning algorithm considered driver individual characteristics less, which makes the applicability of CAS poor. Real-time identification of driver’s driving characteristics and accordingly select the appropriate collision avoidance-warning algorithm or adjust some parameters to make the CAS more correct, which is the development trend of CAS. The driving characteristics due to difference of the individual driver’s physiological and psychological status can be defined as driving tendency. So study the identification of driving tendency real-timely and how can introduce driving tendency to collision avoidance-warning algorithm has a great significance. This paper will have a study on the real-time identification of driving tendency and the modeling longitudinal safe distance. Details are as follows:Firstly, according to the traffic psychology and other existed research achievement analysis the driving tendency characteristics and divide the driver’s type, which can provide a theoretical basis to identify the driving tendency.Secondly, identify the static driving tendency. Design static driving tendency questionnaire which have a high reliability and validity. Then, use the questionnaire to make a test on 128 drivers and conduct factor analysis to the test results and interpret.Thirdly, identify the dynamic driving tendency. Analyze the drivers’ difference between the car-free flow statement and free-driving statement, and build a framework of the dynamic driving tendency’s identification. Then use real vehicle test to collect the driving data of different drivers, and use the mathematical methods of BP neural network, RBF neural network to build the identification model of dynamic driving tendency in the two different statuses and validate the model finally.Fourthly, establish a new safe distance model. Deduce the kinematic safe distance model and improve it by analyzing the automotive braking process. Then through driving simulation experiments to obtain the date of different drivers’ reaction time and braking deceleration 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.