Analysis And Modeling Of Driver Merging Behavior For Intelligent Vehicle Control Strategy

The merging of vehicles is limited by the time and space of the acceleration lane and the main traffic flow state.The driver often shows excellent decision-making effect in the dynamic merging environment.If the driver’s merging process data is modeled and its decision-making mechanism is extracted,it can provide theoretical support for the autonomous merging process of intelligent vehicles.Currently,there are a lot of studies on the driver’s lane changing.Although the merging process is similar to the lane changing process,the conventional lane changing model may not be applied to the merging decision-making process of intelligent vehicles in heavy traffic flow environment due to the remaining length of the acceleration lane and main traffic flow.In view of the above requirements,this paper takes coupe as the platform and uses the lane line identification system,millimeter wave radar,eye tracker and other equipment to build a comprehensive test vehicle for driving behavior.40 test drivers are publicly recruited to carry out a real driving test on expressways.At the end of the experiment,573groups of merging behavior data were screened from the experimental data.The main research contents and conclusions of this paper are as follows:（1）According to the analysis of the relevant characteristic parameters,the vehicle speed does not change much during the merging process,and more than 90%of drivers choose to merge at the beginning and middle of the acceleration lane,and control the speed more often by controlling the accelerator pedal.（2）The driver’s merging behavior is divided into safe merging behavior and dangerous merging behavior,and based on this,the key parameters in the process of merging behavior are analyzed.The results show that the merging duration and line pressing time of the two kinds of merging behavior obey normal distribution.（3）The relative speed,relative distance and the remaining length of the acceleration lane were used as the input,and the driver’s merging decision model was established using the theory of support vector machine.The output was safe merging behavior or dangerous merging behavior.The training set and the test set are randomly assigned in a ratio of 3:1,and the model does not have a high accuracy in the identification of dangerous merging behavior.（4）The training set and the test set are determined according to the data of the dangerous merging behavior,and the model has a high accuracy in identifying the dangerous merging behavior.The relative speed between the merging vehicle M and the₍（9）vehicle,the relative distance and speed between the merging vehicle M and the₍（9） vehicle,and the remaining length of the acceleration lane S are taken as the input parameters to identify again.The results show that the driver’s merging decision-making model can flexibly make decisions based on the input of the model.