Research On Lane Changing Interactive Behavior At Urban Expressway Based On Two-vehicle Driving Simulation

Lane changing behavior is one of the most common driving behaviors,eapically in urban roads,drivers often change their lane for a higher speed and lager space.The interactions generated in this process may affect not only road traffic efficiency,but also safety.At present,research on the lane changing behavior based on a driving simulator is mostly done by conducting a single-vehicle experiment.The movement of other vehicles is preset and will not change with the behavior change of drivers in the experiment.In this case,only the characteristics of one driver can be concerned,and it is difficult to consider the interaction between lane changing vehicle and the vehicle behind it at the target lane.In the real road environments,there is a process of mutual influence and game decision-making in the process of lane changing,and the influence of other vehicles can not be ignored.Therefore,it is particularly important to study the interaction behaviors between two vehicles during lane changing,so as to improve traffic safety and to ensure the stability of traffic flow.Taking interactive driving behavior as the research object,this study obtains the behavior data through a two-vehicle simulated driving experiment,analyzes interactive drivers’ behavior characteristics in the lane-changing process,and explores the influencing factors of their lanechanging decision.After that,a short-term forecasting model of interactive driving behavior based on interactive characteristic parameters is proposed.The specific research is as follows:First,focusing on lane changing behaviors of vehicles in urban expressway scenarios,we introduce environmental factors such as speed limit and vehicle distance,design a simulated driving scenario and carry out a two-driver experiment using the interactive driving simulation system at Southeast University.In this experiment,we collect the interactive driving behavior data of 40 participants,complete the data preprocessing,and obtain the driver’s personalized information based on a driving style questionnaire.Next,based on the theoretical research of lane changing behavior,the lane-changing process is divided into two phases: intention phase and implementation phase.The main characteristic parameters of drivers’ behavior and related influencing factors in the two phases are invesitgated.By comparing the differences of driving behavior characteristics between single vehicle and two vehicles,the necessity of performing a two-vehicle simulated driving experiment is verified.Afterwards,drivers’ lane changing decision combinations are classified according to their acceleration,and a multivariate Logistic regression model is established to explore the influencing factors and its statistical significance.Finally,the decision-making mechanism of drivers in the process of lane changing is analyzed from a micro perspective.On the one hand,by considering the drivers’ behavior parameters and interaction characteristic parameters,an ARIMAX model is built to make a short-term prediction of the driving behavior of the rear vehicle at the target lane.On the other hand,by combining environmental factors and drivers’ characteristics,a LSTM model is established to make a short-term prediction of the acceleration type for the rear vehicle.The effectiveness of the prediction results of the two models is verified through case analyses.By comparing the prediction results of the LSTM model,it is found that the addition of behavior parameters can improve the accuracy of model prediction.