Research On Coordination Method For Single Intersection Based On Cooperative Vehicle-infrastructure Technology

Cooperative vehicle-infrastructure system (CVIS) has become a hot researchfield of the road transportation. While it is the future of the Intelligent Transportation,it is very important to explore how to improve the efficiency of intersection traffic byusing the autonomous vehicles and wireless communication between vehicles andinfrastructures, which are the new features of cooperative vehicle-infrastructuresystem.This dissertation focuses on the efficiency problem of intersection traffic. On thebasis of CVIS, two new intersection coordination methods without traffic lights,which can reduce the delay of vehicles and improve traffic efficiency, have beendesigned. The major research work includes the following aspects:Unsignalized intersection coordinate system framework based on CVIS isresearched. By using the new features such as wireless communication betweenvehicles and infrastructures, an centralized vehicle coordinate framework withouttraffic lights is designed for dual multi-lane intersection.A new stop-and-go-style intersection coordinate method is designed. While thestate of all vehicles in the intersection can be obta ined, a new traffic controller isproposed which removes traffic lights in intersection. The heart of the proposedcontroller is an integer linear program (ILP), which aims to reach the maximum trafficflow. By solving this ILP problem, this method can pro vide a strategy to releasevehicles in intersection. This controller manages safe driving patterns crossing ajunction while it tries to maximize number of passed vehicles across intersection.Simulation results show that, in this new system, the average delay, the average queuelength and the parking rate of vehicles are much lower than that in traditional trafficlights system and some other unsignalized system based on CVIS.A new intersection coordinate method without stop-and-go-style is designed. Byanalyzing the avoiding behavior of human driver in unsignalized low traffic flowintersection, a new intersection coordinate method which based on vehicles trajectoryprediction is designed. This method abandoned the traditional stop-and-go-styletraffic control mode. By converting the vehicle traffic problem into a constrainednonlinear optimization problem (CNOP), this method can defuse the conflict ofvehicles in the intersection without stopping any of them. The solving progress of thisCNOP is also present in this paper. Simulation results show that, in the low traffic flow condition, this method can achieve the non-stopping target, and the coordinateperformance is also much better than the stop-and-go-style method we designed.Intersection traffic simulation platform based on CVIS is modeled and built. Inorder to test and verify the efficiency of the new intersection coordinate methods, anintersection traffic simulation platform based on CVIS is designed by agent modelingidea. The major objects in this platform include vehicle model, road model andintersection controller model. In addition, some evaluation index and otherintersection coordinate methods are added into this platform, just make sure theobjectivity of our test and verify.