Research On Human-like Mandatory Lane Change Strategy In Complex Traffic Environment

Lane changing strategies in complex traffic environments are currently one of the hot topics in intelligent driving research.As a complex traffic behavior,lane changing not only affects the driving efficiency,safety,and comfort of lane changing vehicles,but also affects the operational performance of the entire transportation system as a component of the traffic flow.Frequent and inevitable mandatory lane changing often has a greater impact than free lane changing.Therefore,the study of anthropomorphic mandatory lane change strategies in complex environments is of great significance.Due to the current lack of research on considering anthropomorphic factors in lane change strategies in complex traffic environments,this paper analyzes and studies three aspects of trajectory prediction,lane change decision-making,and lane change trajectory planning in mandatory lane change scenarios.Using trajectory prediction models that consider lane change states,vehicle trajectory prediction is achieved.Based on this,a mandatory lane change decisionmaking method considering anthropomorphic factors such as driving style,lane change pressure,and driver’s field of vision was established.Finally,a set of anthropomorphic mandatory lane change decision-making and planning strategies is constructed based on comprehensive consideration of safety,lane change efficiency,and personalized driver needs,aiming at dynamically generating lane change trajectory based on the characteristics of multi vehicle interaction in complex traffic environments.Firstly,this paper studies trajectory prediction methods considering lane changing states.Based on the commonly used the long short term memory neural network based trajectory prediction model,lane change state judgment is introduced,and the heading angle threshold is used to determine whether the vehicle is in a lane change state.The input of the trajectory prediction model is determined through the lane change state,in order to improve the accuracy of trajectory prediction.The model is trained and analyzed using the NGSIM dataset processed by the two-part reconstruction method based on wavelet transform,which proves the accuracy and robustness of trajectory prediction.Secondly,this paper studies a mandatory lane change decision-making method considering anthropomorphic factors.In order to make the decision-making process more consistent with the behavioral characteristics of real drivers,the lane change decision is divided into three parts: lane change intention triggering,game revenue calculation,and feasibility judgment.Based on the traditional lane change model based on game theory,an intention triggering model based on lane utility is added,and the impact of lane change pressure and driving style is considered,making the mandatory lane change decision more personalized while meeting the personalized needs of different drivers.Thirdly,this paper studies the dynamic lane change trajectory planning method.In order to obtain the optimal lane change path that meets the lane change efficiency,comfort,and different driver needs,this paper selects a quintic polynomial as the lane change path and obtains the optimal lane change path from the candidate set under the premise of considering lane change efficiency and comfort.When generating the speed curve,travel efficiency,comfort,and safety are comprehensively considered,using quadratic programming to obtain the speed curve along the optimal lane change path and assign different objective function weights to different driving scenarios and different types of drivers,fully utilizing the advantages of quadratic programming model in terms of fast solution speed and ensuring the efficiency and comfort indicators of the lane change path.Finally,this article establishes a typical mandatory lane change scenario in VTD,and configures corresponding vehicle dynamics models therein.The model in this article is built in Simulink.The dynamic environment is built through joint simulation to verify that the model proposed in this article can reasonably complete the mandatory lane change behavior and ensure driving efficiency and safety under different driving styles and driving environment changes.