Research Method Of Driver’s Lane Changing Risk Identification

Lane change is a typical driving behavior pattern in the course of driving.Compared with the driving behavior such as car following and free driving,driver’s complicated operation,short decision time and other characteristics are more prominent It has high requirements for the driver’s psychological and physiological qualities.It is a kind of unsteady,easy to distortion and high accident driving behavior.The channel change warning system can remind the driver to adjust the driving behavior or revoke the change of lane when there is a large risk of changing the lane,which can effectively reduce the rate of changing the accident and improve the capacity of the line.However,the research on the key problems of the existing channel early warning system is not thorough enough.There are some problems,such as the deviation of time judgment,the outstanding phenomenon of missing and misreporting,and the shortage of early warning time.Therefore,on the basis of the existing early warning system,this paper puts forward some suggestions on the change of the early warning system and improves the efficiency of the system greatly,striving to improve the accuracy of the system and increase the early warning time.In this paper,we select the male skilled driver,the male unskilled driver,the female skilled driver,the female unskilled driver each 18,carry on the real car test to 72 drivers on the expressway and the city road respectively,and use the GPS system,the millimeter wave radar,the faceLAB5 eye movement instrument and so on to collect the visual search characteristics,the surrounding position relationship and the exercise in the window when the driver is changing the road.The related parameters of characteristics and psychological stress degree provide data support for risk identification of lane changing.The main conclusions are as follows:①The channel intention factor CII and the search time long T are proposed as the selection criteria.By comparing the difference between the lane change intention factor CII and the search time length,the CII threshold and the T threshold are determined when the lane change intention is generated.Thus it is possible to determine the length of the window length of 3S,4.1s,3.6s and 4.6s for male skilled drivers,male unskilled drivers,female skilled drivers and female unskilled drivers.②According to the principle of constructing the index system and the difference of lane changing and lane keeping,the selection index mapping system includes the average time G of the single watch,the average coverage of the single scan S,the standard deviation of the horizontal direction vision,the standard deviation of the head horizontal angle beta,the percentage P of the gaze time of the rearview mirror(1.5 seconds),the car driving speed V,and the auto car.The relative distance of the current Lane front car is TTC1,the relative distance between the car and the target lane front car is D3,the collision time between the car and the target lane front car is TTC3,the relative distance between the auto and the target lane rear car is D4,and the collision time between the car and the target lane rear vehicle is TTC4.③Establish a prediction model of lane changing behavior of GA-SVM.With the comprehensiveness of lane changing behavior index system,the prediction accuracy is increasing.When the input vector contains all the feature indexes,the prediction accuracy is the highest(91.59%),so it is selected as the final input eigenvector.The specificity,sensitivity and accuracy of skilled drivers are higher than those of non proficient drivers,and the rate of false reporting is lower than that of non proficient drivers.According to the time series analysis of the prediction results,it is found that it can identify the change behavior samples quickly and compare the recognition speed with the turning light opening rate.It is found that the GA-SVM model is more accurate and faster than the turning lamp.In addition,the cumulative value of the female driver’s sensitivity was found to be larger than the male driver at[-3.5 and-1.5],indicating that the female driver was more ahead of the male than the male.④To construct the characterization index system of lane change early warning,including horizontal vision standard deviation theta,head horizontal rotation angle standard deviation beta,TTC3 from car and target lane front car,D3 of auto car and target lane front car,TTC4 of self car and target lane rear car,D4 of auto and target car.The 5%and 95%quantiles were identified as the typical evidence values of D-S evidence,and the value of the basic trust probability m(A)of evidence theory was used as the basis of risk assessment grade.⑤From time to time,the risk value of change sample is analyzed from time to time,and it is found that the risk of crossing lane change is greater as time goes ahead.Through the comparison of male skilled driver,male unskilled,female skilled driver and female unskilled driver,it is found that the driver’s proficiency is an important factor affecting the driver’s change of risk.The driver’s change of way is divided into low and middle risk prudent driver,low and middle risk risk driver,high risk and prudent driver,high risk and risk driving driver.The risk degree from high to low is as follows:high risk prudent type,high risk risk,low and middle low risk,middle and low risk,and put forward suggestions to reduce risk of changing road risk by high risk prudent type,high risk risk type,low and middle risk risk driving driver.