| With the advent of the artificial intelligence era and the rapid development of communication technologies,autonomous driving has become an inevitable development trend of future road traffic.However,in the current environment of the internet of vehicles,fully autonomous driving still cannot be applied to people’s daily life.In addition to improving the performance of vehicles and environmental sensing devices,How to achieve safe and efficient automatic driving of vehicles and how to use limited communication resources efficiently are the keys to breaking through the current development dilemma of automatic driving.Intersection is a typical application scenario of the Internet of Vehicles,and it is also difficult to achieve high-reliability autonomous driving in this scenario.A highly robust and efficient scheduling system is required in the process of scheduling and controlling vehicles at intersections.In addition,due to the increasing number of road vehicles,the scheduling system should also have lower complexity to adapt to the current reality of the development of the Internet of Vehicles.In order to deal with the collision and packet loss problem caused by frequent information interactions between the vehicles and the scheduling system and to maintain the stability of the road,the communication system in this scenario needs to consider the time-sensitive performance of vehicle information to avoid outages of the vehicles with the scheduling system when the vehicle controller is noisy.Aiming at the problems of high computational complexity and poor communication freshness of traditional scheduling systems,this paper proposes a dynamic centralized intersection scheduling control method and communication optimization strategies in intersections’ scheduling scenarios.Based on the dynamic characteristics of the vehicles,the information timeliness measurement index in the network,the age of information,is defined.In order to cope with the tr ajectory deviation caused by the dynamic entering and leaving of vehicles and the perception and control noise,the segmented solution method is adopted,and the linear iterative solution of “pipeline” type is adopted to reduce the complexity of scheduling control.In order to avoid the problem that the vehicle status information cannot be connected to the channel and cause the vehicles to lose control,a timeliness-based information transmission interval optimization method and a backoff mechanism improvem ent algorithm based on reinforcement learning are proposed to increase the utilization rate of the communication link and to further improve the robustness and the throughput of the scheduling system.It is verified by simulation experiments that the dynamic centralized intersection scheduling method proposed in this paper can not only improve the scheduling efficiency in high vehicle density scenarios,but also can improve the robustness of scheduling control system.In addition,compared with traditional scheduling methods,this method has lower computational complexity.The communication optimization solution for centralized scheduling scenarios effectively improves the timeliness of information in the system and the utilization of communication link resources,and further improves the throughput performance of intersection scheduling control system. |
