Research On Intelligent Vehicle Lane-change Control Based On Surrounding Vehicle Driving Intention Recognition

Lane changing is one of the most common driving behaviors in vehicle driving,which commonly exists in the process of vehicle driving,especially in the environment of multivehicle driving,unreasonable lane changing behavior is the main cause of traffic accidents,which not only causes traffic congestion,but also threatens the life safety of drivers and passengers.Therefore,how to perform lane change safely in smart cars has become a hot issue for research nowadays.In order to improve the safety and efficiency of smart car lane change,this paper investigates the trajectory prediction method and lane change control method of smart cars in multi-vehicle environment.Among the existing trajectory prediction models,machine learning and deep learning methods are widely used,and long short-term memory network(LSTM)has good performance in long time domain prediction,which has received more and more attention from researchers and scholars.Inspired by this,this paper proposes an intention recognition-based trajectory prediction method(GMM-LSTM)for predicting the driving trajectory of the surrounding vehicles on top of the existing LSTM methods.Among them,Gaussian mixture model(GMM)is used to identify the driving intention of the vehicle,and the corresponding LSTM network is used for trajectory prediction according to the identified results.Then,a driving simulation platform is built,and the model is trained and tested based on the experimental data obtained from the simulation platform,and compared with the existing methods.The experimental results show that the proposed model has good prediction accuracy,compared with the traditional CA model and the LSTM model without intention recognition,the prediction accuracy is improved by 45.9% and 17.3%,respectively.Secondly,the intelligent vehicle lane change decision and trajectory planning methods are studied.By analyzing the factors affecting the vehicle lane change,the lane change decision logic of the intelligent vehicle is established,and the lane change decision model is built based on the Matlab fuzzy logic toolbox,the quintuple polynomial curve with minimum jerk is selected as the intelligent vehicle lane change trajectory planning model,and the lateral acceleration and trajectory curvature generated by the lane change are considered,while taking into account the safety of the intelligent vehicle lane change,the cluster family of lane change trajectories is established The optimal lane change trajectory is selected by considering the transverse acceleration and trajectory curvature generated by lane change,and taking into account the safety of lane change,the evaluation function of lane change trajectory cluster family is established.Finally,an MPC trajectory tracking controller was designed based on a joint simulation platform to accurately track the planned lane changing trajectory for intelligent vehicle lane changing control problems.Based on the constructed vehicle model,a predictive model of the MPC controller was established.Based on actual control problems,the constraint conditions and objective functions of the controller were designed.A joint simulation experiment of lane change control was conducted in Matlab Simulink and Carsim to track the lane change trajectory planned in the upper layer and verify the effectiveness of the control method.