Research On Decision Control Of Automated Driving Vehicles On Freeways Under Connected And Automated Environment

With the rapid development of new technologies,such as big data,artificial 5G communication and intelligence,Intelligent Traffic System(ITS)gradually upgrades from traditional stage to “internet of mobility” stage,the new generation ITS with automated driving as a core becomes the current research hotspot.Since the driving environment on freeways is relatively closed and stable,and the freeways provide a landing path for the application of automated driving technology,based on the better foundations of software and hardware equipments.As a novel automated driving technology,connected and automated vehicle(CAV)has drawn extensive attentions in recent years,and has made some progress in theoretical and practical applications,such as most automobile companies currently using the automation control systems as the solutions of automated driving.The essence of the solutions is that the vehicle itself can achieve automated driving to some extent,with no assistances and partial assistances from road infrastructures(i.e.traffic control center).However,current few studies focus on the key issues of automated driving on preliminary stage of internet of mobility where the communication technology cannot be large-scale deployed.The complexity of decision-making will be increased,due to the state information of automated vehicle(AV)cannot be collected,which severely restricts the applications of CAV.However,freeway systems can achieve large-scope sensing,communication and decision-making on the preliminary stage.Under the development trend,vehicle-road infrastructure(vehicle-infrastructure)collaboration automated driving technology can provide a new way to solve the mixed traffic flow problem of automated driving vehicles.The study considers the mixed traffic flow of automated driving vehicles as the research object in an environment of internet of mobility,including CAV,AV and CAV platoon(CAVP).The compositions and application architecture of the automated driving-based traffic system in an environment of internet of mobility are introduced.Combining the operational advantages of vehicles and road infrastructures,based on the feedback control and optimal control algorithms,the vehicle-infrastructure collaboration decision-based car following and lane change method of automated driving vehicles are proposed.Considering the features of vehicle degradation/upgradation in a CAVP,a new distributed control system of CAVP is designed to cover the switching function between AV and CAVP driving mode.Analyzing the features of communication topology and control mode,the vehicleinfrastructure collaboration decision-based distributed and centralized control method of CAVP are formulated,thereby providing multiple solutions for the practical applications of CAVP.The main content and results of the study include the following aspects:First,the study focuses on the composition and application architecture of automated driving-based traffic system in an environment of internet of mobility,and consider vehicle-infrastructure collaboration automated driving technology as a solution with the upgrading role of the road.Firstly,three key sub-systems of vehicle,infrastructure and communication are described respectively,and three major functions of sensing,decision-making and control are also presented.According to the definition for driving automation proposed by society of automotive engineers international,the study proposes the taxonomy and definition of road infrastructure-based automated driving,including six stages from non-informatization,non-intelligence,nonautomation(I0)to fully automated driving based on road(I5).Finally,the technical levels and trends of both vehicles and infrastructures are analyzed,the environment connotation and application process on the preliminary stage are then determined.Second,the vehicle-infrastructure collaboration decision-based car following control method of automated driving vehicles is proposed.The study taking the traffic throughput as the optimization variable and the string stability as the constraint,the control method is implemented from the aspects of control mode and information mode.The control mode is divided into strict and loose,and the information mode is divided into Vehicle to Infrastructure(V2I)and V2I+Variable Message Board(VMS).Theoretical analysis and simulation experiments of multiple mixed traffic scenarios are conducted.The results show that relaxing the string stability conditions can enable collaborative control between CAVs,thereby improving traffic throughput and ensuring overall head-to-tail stability.For the implementation of V2I+VMS,it can promote AVs to execute control instructions and then increase the number of control vehicles,thereby provide more effective solutions to improve traffic operation performance.Third,the vehicle-infrastructure collaboration decision-based lane change control method of automated driving vehicles is proposed.It is suitable to the mixed traffic scenarios where the communication technology is not deployed on a large scale.Based on the perceived state information and predictive control parameters of the surrounding vehicles by road infrastructure,a cooperative lane change control method is formulated from the safety,efficiency and comfort aspect,and make a comparison with the noncooperative control method.The reachability set of vehicle trajectories under uncertain sensing conditions are analyzed,and total running costs using the proposed method are then calcualted to adjust the automated driving decisions.The results show that the proposed control method by combining the advantages of vehicle and infrastructure can provide the optimal decisions and improve the success rate of lane change.Fourth,with respect to vehicle degradation/upgrade scenarios arising from the vehicle cut-in or communication interruption,a new distributed control system of CAVP is proposed to cover the switching function between AV and CAVP driving mode.The stability theory is used to verify the effectiveness of the proposed control system.The results show that the proposed control system can use a smaller desired time gap to guarantee the string stability compared to the Plog’s control system.Through traffic simulation experiments,the results demonstrate that the proposed control system can effectively improve traffic operation and safety.Considering the differences of decision maker,the vehicle-infrastructure collaboration decision-based distributed control method of CAVP is proposed,and the impacts of the two control methods on traffic operation and safety then analyzed.It can provide the scientific support for the optimal design of CAVP distributed control system.Fifth,the vehicle-infrastructure collaboration decision-based centralized control method of CAVP is proposed.Based on the perceived state information and predictive control parameters of the surrounding vehicles by road infrastructure,the state space and control vector of CAVP contrl system can be re-detemined to formulate a CAVP centralized control method,which is suitable in the scenarios of the mixed traffic flow of automated driving vehicles.Simulation experiments under lane drop or closed are conducted to evaluate the effectiveness of the proposed control method.The results show that the vehicle-infrastructure collaboration decision-based control method can effectively reduce the total travel time and collision risks compared to vehicle decisionbased method,and the cooperative control mode performs better than the noncooperative control mode in traffic operation and safety.Moreover,the vehicle group of lane change including CAV has a better traffic operation and safety performance.