Research On Intelligent Vehicle Trajectory Planning And Tracking Method Considering Trajectory Interaction

Intelligent vehicles are a comprehensive system that integrates functions such as environmental perception,decision planning,and control execution.Planning and control,as the core link of intelligent vehicles,are important guarantees for achieving safe driving of vehicles in complex traffic scenarios.In traffic scenarios,the motion of all vehicles is uncertain,so the planning system of intelligent vehicles needs to consider the trajectory interaction between vehicles in the scene.In response to this issue,this study focuses on three aspects:vehicle trajectory prediction,trajectory planning,and trajectory tracking control.The specific content is as follows:1.This article designs a vehicle trajectory prediction model that combines driving intention recognition.Firstly,the random forest algorithm is used to complete the task of vehicle driving intention recognition,and Bayesian optimization is introduced to find the optimal hyperparameter combination;Secondly,in response to the weak feature mining ability of longterm and short-term memory networks,a convolutional neural network is added to construct a trajectory prediction model,and vehicle trajectory prediction is performed based on the results of driving intention recognition.Finally,quantitative and visual analysis was conducted on the prediction results,proving that the prediction model proposed in this paper has higher accuracy.2.In response to the fact that most trajectory planning methods do not consider the interaction between vehicles,this paper proposes a trajectory planning method that considers vehicle trajectory interaction.In order to reduce the complexity of the planning problem,the trajectory is dimensionally reduced.Firstly,by comparing and analyzing some physical properties of commonly used planning curves,quintic polynomial curve is selected to generate a trajectory set,and a multi-index evaluation function is designed to screen the optimal trajectory;Secondly,the track prediction results of surrounding vehicles are mapped to the ST diagram,and the dynamic programming method is used to search the optimal speed curve,and the Quadratic programming method is used to smooth the speed curve.Through comparative experiments,it has been proven that the planning method proposed in this article can plan more reasonable trajectories.3.In order to reduce the computational complexity of the tracking algorithm,this paper decouples the lateral and longitudinal motion for separate control.Firstly,an improved linear quadratic regulation control algorithm is proposed for lateral control,introducing feedforward control to compensate for static errors,and designing a weight dynamic tuning rule to improve the adaptability of the control algorithm;Secondly,for longitudinal control,considering that traditional model predictive control algorithms can only track speed and cannot accurately track longitudinal distance,an improved model predictive control algorithm is proposed,taking into account vehicle’s position and speed errors,and combining multiple objective functions to achieve speed control of the vehicle.Through testing under typical operating conditions,it has been proven that the tracking control algorithm proposed in this paper performs better in terms of tracking accuracy.4.To demonstrate the effectiveness and reliability of the above algorithms,a joint simulation platform was established to simulate and analyze the planning and control algorithm designed in this paper in typical scenarios.The results showed that the planning algorithm considering vehicle trajectory interaction can achieve reasonable trajectory and speed planning,and the improved tracking algorithm can achieve accurate tracking.