Modeling And Simulation Of The Lane-changing Execution Behavior Of Autonomous Vehicle

In recent years,with the development of Internet technology and the improvement of the automation level,Automatic driving vehicles are becoming more and more concerned by Internet companies and traditional car companies,Autopilot technology is beneficial to society,drivers and pedestrians,which can reduce the incidence of overall traffic accidents,alleviate traffic congestion and traffic pollution for its more efficient driving mode.It is considered to be an important method to solve the traffic problems in the future.Therefore,it has also attracted the attention of many scholars in the field of transportation.Lane-changing behavior is not only the key issue in the field of automatic driving,but also the key issue in road basic driving behavior,which has a significant effect on the traffic flow characteristic and the traffic safety.The main research contents include two parts: the lane-changing decision and the lane-changing path planning.However,compared with the cruise control technology of the field of automatic driving,vehicle lane-changing behavior automation has just started.In the existing academic research papers,on the one hand,most of the content change lane decision only pay attention to the free lane-changing behavior of the vehicle in a straight road scene,but for the mandatory lane-changing behavior in the ramp-off scenario of freeway,few models can quantify the relationship between the lane-changing intention point and the lane-changing success probability.On the other hand,most of the existing lane-changing path-planning models cannot be applied to the real driving environment,which is a static lane-changing path planning(SLPP).However,the dynamic lane-changing path planning(DLPP),which can respond to the change characteristics of environmental information in real time,is still limited.This paper tries to fill the two parts of the autopilot lane changing field.In this paper,our research direction is the modeling and simulation of the lane-changing behavior under the background of automatic driving.Firstly,in order to solve the problem of lane-changing intention generation in the field of automatic driving,a lane-changing intention generation model is established,which quantifies the relationship between success probability of lane-changing behavior and lane-changing intention generation point.Secondly,we build an autopilot path-planning model.The autopilot lane-changing path-planning model is modeled to achieve the dynamic path-planning response in the process of lane-changing process,where we add a lane change rollover algorithm,a collision avoidance algorithm and restructure the speed changing rule.It comes out a compete autopilot dynamic path-planning model which can achieve the adjustment of reaction time,planning step.Meanwhile,we use the NGSIM data as the driving information input to simulate and induce our simulation result to four typical lane-changing scenarios,finally,we validate the executability of the planning trajectory in the CarSim environment.The lane-changing intention generation model is a macro discrete probability model of automatic driving,which quantifies the relationship between success probability of lane-changing behavior and lane-changing intention generation point and it is conducted to support for autopilot lane-changing intention generation.The results show that our automatic driving lane-changing path-planning model can be applied to the following several typical scenarios: overtaking to the slower lane overtaking,overtaken to the faster lane,and returning to the original lane.CarSim simulation results show that the trajectory and speed of the model can be tracked well,and the output results of all vehicle parameters are normal.The relative error between lane-changing intention generation model and real vehicle lane-changing experiment result is about 10% and it can be used in the decision optimization of lane-changing intention generation.