Research On Autonomous Lane Changing Method For Intelligent Vehicles In High-speed Environment

In recent years,the number of cars in the country has gradually increased,which poses great challenges to transportation safety and efficiency.Intelligent vehicles,equipped with sensors,can replace human drivers to complete driving behavior in the process of sensing,decision-making and control,which can greatly reduce the accident rate To ease traffic pressure.The realization of safe and efficient autonomous lane change is a key step for intelligent cars.In this paper,through the design of data acquisition platform with advanced sensors,the actual driving scene data acquisition,through data processing and labeling,complete the extraction of lane changing scene,provide data support for lane change research.By defining the lane change process,summarizing the reasons for the success and failure of lane change,statistics the number of scenes according to different reasons,simplify the scene of lane change,and analyze the changes of vehicle parameters in the process of lane change.Establish the safe distance model of lane change to ensure that the vehicles do not collide.Due to the subjectivity of the lane change process,the fuzzy control theory is used to control the lane change,and the left lane change scene is constructed.Two indexes of relative speed and relative distance are introduced as the decision conditions of lane change.The rules of lane change process are extracted from the data.The effectiveness of the method is verified by comparing with the actual lane change scene.Next,the trajectory of lane change is selected.By fitting the real data points,the track shape similar to the actual track is selected.The track constraint conditions are established through the definition of the lane change.Different trajectories are compared from multiple dimensions.The polynomial trajectory is improved by setting the evaluation function,and the optimal trajectory is obtained,which takes into account the comfort and efficiency of the lane change.The lane change trajectory of different driving styles are compared.The turning radius constraint is established to ensure the feasibility of lane change under the limit condition.Finally,the trajectory tracking control is completed using the model control theory.The vehicle dynamics model and tire model are built,and the non-linear dynamic model is simplified through reasonable assumptions.Because the nonlinear model is complicated and computationally intensive,through linearization and discretization,the constraint conditions in the control process are established,and a linear discretized model predictive controller is constructed.Finally,in the CarSim and Simulink environments,based on different drivers The results show that the method is very robust under different vehicle speeds,different lane changing times,and different road friction coefficients.