Research On Intelligent Vehicle Trajectory Planning Method For Lane Change Based On Vehicle-infrastructure Cooperative System

With the development of automobiles to be intelligent and connected,the traditional single vehicles are facing with the problems of limited sensing range and high cost of vehicle sensors.The vehicle-infrastructure cooperative technology has been gradually promoted and applied in the field of automatic driving.Through the real-time interaction of dynamic information between vehicles and infrastructures,it can provide more comprehensive environment-aware information for vehicles.As an important working condition for intelligent vehicles,vehicle lane changing behavior is one of the important indicators of automatic driving test and evaluation.The vehicle lane change based on the vehicle-infrastructure cooperative technology can improve the perception of the surrounding environment and the safety level.Therefore,based on the vehicle-infrastructure cooperative technology,this paper studies the intelligent vehicle trajectory planning method in the lane-change typical scenario.The fusion decision of road perception and vehicle perception is an important issue.Establishing an appropriate decision-making architecture is the key to vehicle-road collaboration.This paper designs a vehicle-road collaborative decision-level fusion architecture.A two-layer trajectory safety verification strategy is proposed based on vehicle-infrastructure cooperative system perception and the vehicle-mounted perception.Then,design the vehicleroad communication module,including the selection of communication schemes and data interaction protocol.The vehicle terminal program development is completed based on the TCP protocol and the ROS robot operating system under Linux,and the function is tested indoors.Communication delay exists in vehicle trajectory planning based on vehicle-infrastructure cooperative system.Asynchrony and polycentricity of roadside and vehicle perception information lead to multiple stages of the control process.Therefore,GAUSS pseudo-spectral method is used to carry out trajectory planning,which has good ability to solve multi-stage problems.Thus improving the robustness of the system.The multi-stage optimal control model for path planning is established based on traffic scenarios.The P-Norm function is used to standardize the vehicle-road data in multiple formats.Uniformly characterize static and dynamic obstacles and various types of roads.Establish vehicle dynamics models,path constraints,and boundary constraints.Lagrange interpolation is used to discretize the optimization problem at the Legendre-Gauss(LG)collocation point.The optimal control problem is transformed into nonlinear programming problem by orthogonal polynomials approximate differential equations,and solved by sequential quadratic programming method.A model-in-the-loop simulation platform based on Prescan software is set up to perform trajectory planning on the lane changing scene.Secondly verify the trajectory planning result,and then use the MPC trajectory tracker to track the trajectory.The trajectory has good followability and anti-interference ability.A vehicle experiment was carried out in Guangzhou International Biological Island to verify the trajectory planning algorithm based on vehicleinfrastructure cooperative system.The simulation results and experimental results show that the vehicle-road collaborative decision-making architecture and trajectory planning method designed in this paper have good solution effect.Have certain research significance and the application value.