Research On Active Lane Changing Decision And Trajectory Planning Methods For Intelligent Automobile On The Freeway

The autonomous lane change technology of intelligent vehicles is one of the research highlights,and its development is crucial to improving the driving safety of intelligent vehicles.According to the accident statistics of various countries and regions,the accident proportion caused by lane changes accounts for 5%to 10%of the total accidents,seriously threatening personnel and property safety.Therefore,it is important to establish an autonomous lane change decision model and an autonomous lane change trajectory planning model.In order to optimize the safety,comfort,and traffic efficiency of active lane changing by autonomous vehicles,a lane-changing decision model based on the GSCV-LightGBM algorithm and an autonomous lane-changing trajectory planning model based on quintic polynomials are proposed.Firstly,regarding the autonomous lane-changing decision model,this paper uses a simple moving average filtering technique to remove outliers from the public NGSIM dataset.Furthermore,through the analysis of the curvature rate of change of the vehicle trajectory and the analysis of the maximum lateral speed of the vehicle,after determining the start and end position of the lane change,the vehicle free lane change trajectory data is extracted,and the decision feature variables are set to construct the data set.At the same time,the parameters of the LightGBM model were tuned using a grid search cross-validation algorithm,resulting in the GSCV-LightGBM autonomous lane change decision model.The model performs well with high accuracy and a short decision time.Further comparison with other machine learning algorithms shows that the GSCV-LightGBM model has the highest accuracy and shortest decision time.This paper also validates the decision model using the HighD dataset from the University of Aachen.The experimental results demonstrate that the model has good generalization capability,high accuracy,and a short decision time.Secondly,in terms of the autonomous lane-changing trajectory planning model,this paper designs a trajectory planning model based on five polynomial multi-boundary constraints.The model classifies vehicles’ free lane changing behaviour into unimpeded free lane changing,and free lane change with obstacles based on safety distance threshold,sets the local path planning evaluation function,respectively.The jerk is used to evaluate the lane-changing trajectory twice to ensure the planned trajectory’s safety,comfort and traffic efficiency.The multiple boundary conditions include lateral speed and acceleration,longitudinal speed and acceleration,lateral displacement,local path collision avoidance,lane change comfort,and traffic efficiency constraints.Finally,this paper uses a model predictive control approach that controls the simulated vehicle to follow the planned trajectory of the lane change through a joint CarSim-Simulink simulation.Verify the reliability of the lane change decision model and the reliability of the trajectory planning model.Analyzing the front wheel angle,longitudinal velocity,and lateral acceleration information collected during the lane change process demonstrates that the planned trajectory meets multiple constraints and can effectively improve the autonomous vehicle’s comfort,lane change efficiency,and safety.In summary,the autonomous lane change decision model and the trajectory planning model proposed in this paper can optimize the lane change efficiency,comfort,and safety of autonomous vehicles,which positively contributes to the development of autonomous vehicle technology and has certain theoretical and practical implications significance.