Research On Vehicle-in-the-Loop Intelligent Connected Vehicle Simulation System Based On Real-Time Traffic Flow

Autonomous driving testing technology is an important guarantee for the large-scale commercialization of autonomous vehicles.The existing test methods are mainly based on real roads and simulations.Traditional road testing has a real road test environment,but the diversity of test scenes is limited,making it impossible to customize corner case scenes safely and efficiently.Simulation testing is flexible and efficient,but the lack of a real traffic flow test environment separates the strong coupling relationship between the vehicle and the environment in practical application scenes.In view of above,this thesis focuses on the Vehicle-in-the-Loop（Vi L）simulation system based on vehicle-road-cloud collaboration,conducts researches on the road-end perception algorithm,the traffic flow simulation scene enhancement method and the Vi L twin simulation system,and constructs the Vi L intelligent connected vehicle simulation system based on real-time traffic flow.Moreover,in the campus road scene,the autonomous driving decision algorithm is used as the test case to complete the system test and verification in the virtualreal twin traffic flow scene.The research contents are as follows:（1）A method for edge perception of real-time traffic flow data on roads.Aiming at the problem that the traditional simulation test environment cannot map real-time traffic flow,this thesis proposes a "road-cloud" co-simulation method based on road-end edge perception,in which the acquisition of real-time traffic flow data is the basis for constructing a simulation test scene.Based on the advantage of the fixed pose of the roadend camera,this thesis uses YOLOv5 and Deep Sort to complete multi-target tracking,and uses perspective transformation to complete the rough localization of traffic flow elements.Focusing on vehicle elements,this thesis proposes a method for estimating vehicle pose and attitude based on geometric attributes,and achieves precise positioning results at the centimeter level,then lays a realistic foundation for enhanced simulation of traffic flow scenes.（2）Simulation scene enhancement method based on real-time traffic flow.Aiming at how to apply the advantages of rich scenes in the virtual environment to the autonomous driving simulation test system to improve the test efficiency,this thesis opens the data interface in the self-developed simulation platform based on the static scene about roads and buildings,and obtains the road-end dynamic traffic flow data in real time.Consequently,the simulation platform completes the rendering and control of the traffic flow model and realizes the twin simulation of the traffic scene.Besides,this thesis enhances environmental factors like the weather and lighting of the static scenes by selfcustomization,as well as traffic flow elements like the virtual pedestrians and vehicles of the dynamic scenes according to the test requirements,and completes the customization of corner case scenes through the virtual-real twinning,so as to maximize the advantages of "virtual-real" symbiosis,then provide a variety of test scenes for the Vi L system.（3）Construction and verification of a real Vi L twin simulation system for the intelligent connected vehicle.With consideration for how to connect the intelligent connected test vehicle to the aforementioned simulation system to realize the whole vehicle-level system test,this thesis uses the Vi L method to twin the vehicle status data and road-end traffic flow perception results in real time to the virtual scene of the simulation platform.The decision and planning results of the simulation cloud platform are synchronously fed back to the intelligent connected real vehicle test platform to realize closed-loop vehicle-road-cloud co-simulation.This thesis takes the autonomous driving decision algorithm in the campus road scene as an example,then completes the system test in the customized corner case scene.In summary,this thesis builds a closed-loop system consisting of the intelligent connected vehicle and a hybrid test environment.By introducing real-time traffic flow information,it can quickly generate test scenes which conform to real traffic conditions,and realize customizable functions for complex scenes,then provide strong support for the transition from simulation tests to road tests.