Research On Driver's Lane-changing Intention Identification On Urban Freeway

Congestion on the urban expressway is a difficult traffic problem in many large cities,congested driving environment will affect the driver's driving operations and even lead to offensive driving behavior.Forcible lane-changing is a common type of offensive driving behavior that seriously affects traffic safety.If lane-changing intention is identified before the operation,the safety of lane-changing can be evaluated in advance and an early warning can be triggered in time.Now,the researches on lane-changing intention identification mostly set the scenario to the unblocked state without considering congestion.In addition,there is no research on urban expressway scenarios.Based on the summaries of domestic and foreign research results,this paper studies the identification of drivers' lane-changing intention under congested urban expressway condition.Firstly,the lane-changing behavior was analyzed from the perspective of lane-changing type,lane-changing decision mechanism,and lane-changing phase.On this basis,a two-stage real vehicle test program was designed from the aspects of test equipments,test personnels,test routes,and test procedures.Through the test data obtained,the width of lane-changing intention window under different congestion states was determined,and the lane-changing intention samples and lane-keeping samples were screened,providing data support for lane change intention identification.Secondly,driving style surveys was conducted with self-reported questionnaire and a driving style database was thus established.The driving styles were quantified with the principal component analysis method and were classified into three categories with the Kmeans clustering analysis method.Then,the test personnels' driving style can be extracted from the database.Based on the data from the first survey,using driving style and speed as influencing factors influencing the congestion intensity of road segments based on the driver's perception,a road congestion intensity model based on multiple Logit can be established.Thirdly,this paper analyzed the eye movement characteristics,heart rate characteristics,and basic parameters of the vehicle's movement state in the process of driving,such as average fixation time,number of rearview mirror fixations,saccade amplitude,saccade speed,heart rate growth rate,standard deviation of speed,standard deviation of acceleration,standard deviation of yaw rate under the condition of congestion state of unimpeded,slightly congested and moderately congested traffic.Using the significance test,the differences in various parameters of the lane keeping phase and the lane change intention phase under various traffic conditions were analyzed,and the parameters with significant differences were extracted as indicators for identifying the lane change intention.Lastly,based on the support vector machine theory,the feature parameters representing the lane change intention were imported into the model.According to the cross validation idea,the grid search method and the particle swarm algorithm were used to perform the optimization of the penalty parameter C and the kernel function parameter g of the model.The test set was used to test the model.The accuracy rates of the unblocked group,lightly congested group,and moderately congested group were respectively 93.617%,91.304%,and 93.333%,which met the requirements of the lane-changing assist system.Finally,combined with the research results of this paper,the development ideas of lane-changing inte warning system based on congestion status were proposed.