Vehicle-Infrastructure Cooperative Decision Making And Simulation System Implementation

V2X(Vehicle-to-Everything)technology enables all-around connectivity as well as efficient and accurate information communication between vehicles and surrounding vehicles,people,transportation infrastructure,and network/cloud.On this basis,vehicle-infrastructure cooperation achieves intelligent cooperative sensing,decision making,and control through the interaction and Vehicle-Vehicle/Vehicle-Infrastructure information sharing.Vehicle-infrastructure cooperative decision-making determines driving behavior based on cooperative sensing,aiming to improve driving safety,traffic efficiency,and conserve energy.This thesis focuses on the implementation of vehicle-infrastructure cooperative simulation platform,as well as the vehicle-infrastructure cooperative decision making approach for the on-ramp merging scenario.Research on vehicle-infrastructure cooperative decision making is applied to various scenarios,such as vehicle formation,lane changing,intersections,on-ramp merging,etc.The research on vehicle-infrastructure cooperation decision making for on-ramp merging scenarios can be divided into the traditional optimization algorithm and artificial intelligence algorithms.Most research are based on the premise of vehicleinfrastructure cooperation to obtain surrounding information under the condition of perfect communication,and do not take into account the effects of fluctuation of real-time communication performance of network.Research requires simulation to verify the effectiveness of the method Vehicle-infrastructure cooperative simulation is realized through the combination of traffic communication layer simulation software and information communication simulation software.However,the existing platform and software suffer from issues with functional integrity,usability,and persistent data storage.The major work of this thesis includes two aspects:simulation platform and decision making for on-ramp merging based on vehicleinfrastructure cooperation.The main contributions include:A.In order to deal with the problems of the above-mentioned software related to vehicle-infrastructure cooperative simulation,this thesis designs and implements a platform for vehicle-infrastructure cooperative simulation,integrating V2X communications,driving decision/control and traffic simulation functions.In addition,the simulation platform has the following features:a Web interface,simulation management,persistent data storage,and the ability to integrates communication,decision making,and transportation.Users can access this platform via an online web interface or an offline download to conduct tasks such as managing simulation and training data,simulating traffic and vehicle networks,and simulating combined vehicle and road networks.B.This thesis presents an on-ramp merging decision-making method that adapts to communication performance and uses the environmental obtained by vehicle-infrastructure cooperation to make driving decisions.In particular,the merging vehicles in the on-ramp acquire communication performance information as well as the position,speed,and acceleration information of the vehicles on the main road in real time,and use the reinforcement learning-based method to make driving decisions based on the acquired information in order to complete the on-ramp merging process.Finally,algorithm training and the evaluation of experimental results through the vehicle-infrastructure cooperation simulation platform of this thesis that shows how the algorithm improves traffic safety and passenger comfort while maintaining traffic efficiency.Additionally,it confirms the usability and dependability of the platform.