Research On Cooperative Vehicle Control Algorithm For Unsignalized Intersection Under Connected Vehicles Environment

The unsignalized intersection is a concentrated area of traffic congestion and vehicle accidents,and the smoothness of traffic at unsignalized intersections directly affects the efficiency of the entire road network.With rapid development of advanced technologies such as automated driving and connected vehicles,cooperation optimization and control of road intersections has become a critical research topic in the field of transportation.As the major part in this subject,cooperative control between multi-vehicles has great practical significance and broad development prospects.In addition,different vehicles being mixed in traffic will become normal,which puts forward higher requirements for cooperative control mechanism.This Ph.D.dissertation is based on the project of National Key Research and Development Plan named Research on Coupling Mechanism and Cooperative Optimization Method of Cooperative Vehicle Infrastructure System(No.2018YFB1600500).This paper takes connected and automated vehicles and human driven vehicles as research objects,driving safety field theory as the basis,reducing driving risk of vehicles as well as improving traffic efficiency as the aims,and researches on cooperative control of connected and automated vehicles,recognition of drivers’ operational intention at unsignalized intersections and cooperative control between connected and automated vehicles and human driven vehicles in mixed traffic are carrie out in the dissertation and a coupled simulation platform is established to analyze the impact of wireless communication performance on the algorithm.In summary,the research content of the paper mainly includes:The MPC-based driving risk minimization algorithm(MPC-DRMA)is proposed aiming at connected and automated vehicles at unsignalized intersections based on model prediction control and driving safety field theory.The previous cooperative vehicle control algorithms ignore the impact of various traffic participation elements,resulting in certain limitations.In order to better solve this problem,the MPC-DRMA is proposed to deal with the impact of all traffic participation elements on driving safety in human-vehicle-road closed-loop system comprehensively and systematically.With driving smoothness,comfort and driving risk minimization as the control objects,the algorithm reduces the total driving risk of vehicles coming from conflicting paths,and allocates optimal driving strategies for each vehicle approaching to the intersection.The simulation test platform is built based on VISSIM and MATLAB.The results show that the MPC-DRMA algorithm can effectively improve the traffic capacity of unsignalized intersections,reduce fuel consumption and vehicle exhaust emissions as well as reducing driving risk of vehicles.For human driven vehicles at unsignalized intersections,a driver operational intention prediction model based on driver’s perception-decision-behavior is proposed.At present,there are few studies on the coupling mechanism of various elements of cooperative vehicle infrasturcture system,and the driver’s perception mechanism and behavior characteristics are particularly critical.In this paper,based on driver’s risk perception level,driver behavior and vehicle dynamics,a human-vehicle loose coupling model based on hybrid state system(HSS)is constructed,and driver decision-making and vehicle operating state are modelled as discrete state system(DSS)and continuous state system(CSS)respectively.Taking going straight,turning left,turning right and stop typical driving behaviors as research objects,a driver operational intention prediction model based on driver perception-decision-behavior is established by means of gaussian mixture-hidden markov model(GM-HMM),and then the HSS + GM-HMM architecture is designed to estimate driver’s operational behavior at unsignalized intersections,and provides a basis for decisionmaking for establishing a perfect human-vehicle coupling system.Compared with KNN estimation and human estimation,the HSS + GM-HMM framework shows a better estimation result.Aiming at connected and automated vehicles and human driven vehicles at unsignalized intersections,the MPC-based modified driving risk minimization algorithm(MPC-mDRMA)is proposed.As automated driving and connected vehicles technologies gradually mature,it will become the normal for mixed traffic driving environment including human driven vehicles,connected human driven vehicles,nonconnected automated vehicles and connected and automated vehicles et al.Exploring the coupling mechanism of the human-vehicle-road system is very necessary to realize cooperative vehicle infrastructure control in a mixed traffic environment.Therefore,based on the MPC-DRMA algorithm,the two-state safe-speed model and heterogeneous traffic flow model are applied to model human driven vehicles and connected and automated vehicles respectively.The driver’s behavior field is introduced into driving safety field model of optimization objective function.Meanwhile,taking into account that the presence of driver factor will inevitably affect the vehicle speed,the branch to bound algorithm is utilized to group vehicles,furtherly optimize the speed of human driven vehicles.The connected and automated vehicle uses HSS+GM-HMM architecture to estimate driver’s operational behavior at the intersection and interacts with human driven vehicle.The simulation results indicate that the proposed MPC-mDRMA algorithm can effectively improve the driving risk of vehicles whiling crossing the intersection,and show better performance with the increase of market penetration rate of connected and automated vehicles.Under real road scenarios,the impact of communication delay and packet delivery rate on cooperative vehicle control algorithm is analyzed.At present,most researches on vehicle coordination at unsignalized intersections assume that wireless communication environment is ideal and perfect,which is difficult to truly reflect the objective environment.There is great uncertainty in propagation process of wireless communication,complex network topology and frequent channel access requests will lead to problems such as packet loss and communication delay,resulting in instability at linear and nonlinear levels,and the performance of cooperative vehicle control algorithm significantly dropped.The network simulator NS3 is supplemented to the previous simulation platform,and a new coupling simulation platform based on VISSIM、NS3 and MATLAB is constructed.The packet delivery rate and communication delay are used as evaluation indexes to analyze the relationship between the two and velocity of vehicle,number of vehicle and distance between vehicles.In addition,the times of conflict and average speed are selected as indicators to study the impact of communication delay and packet delivery rate on the cooperative vehicle control algorithm,and puts forward some reasonable suggestions based on the analysis results.The research mentioned above,aimes to solve cooperative optimization problem of vehicles at unsignalized intersections,introduces driving safety field theory into multi-vehicle cooperative control strategy to quantitatively describe the driving risk of vehicles.Based on the driver’s risk perception level,the operational intention of human driven vehicle at unsignalized intersections is estimated.The proposed MPC-DRMA and MPC-mDRMA algorithms can improve the performance of cooperative control between different attribute vehicles,as well as traffic safety and efficiency at unsignalized intersections in connected and automated driving environment and mixed traffic environment.