Research On Passenger Car Rear-End Collision Avoidance System Based On In-Depth Investigation Of Traffic Accidents

When autonomous vehicles will fully enter the market is a question that has faced society and the autonomous vehicles industry for many years.The market and government will not accept the full entry of autonomous vehicles into the market until the safety of autonomous driving is proven.All parties have invested a lot of economic and time costs to verify the safety of autonomous vehicles.Scenario-based testing is one of the main methods that simulate traffic situations in a virtual simulation way.Since there are infinite scenarios in natural traffic flow,decomposing these scenarios into limited ones and finding representative ones is the main problem based on scenario testing.Passenger car rear-end collisions are a frequent type of accident in the real world.This thesis relies on the China Travel Safety In-depth StudyTraffic Accident Survey jointly initiated by Central South University and Volkswagen(China)to carry out in-depth accident investigation,accident reconstruction,and safety testing research of autonomous vehicles in rearend collision scenarios.The main research contents of this thesis include:(1)In-depth investigation of passenger car rear-end collisionsThe in-depth investigation system of passenger car rear-end collisions takes on-site data,video surveillance/driving recorder as the original data,and integrates the medical information of casualties assisted by the police and the vehicle damage survey information of the investigation team to truly restore the accident scene and conduct in-depth analysis of the cause of the accident.Extract key characteristics of accidents,build a deep accident database,and then provide critical data support for research on accident injury mechanisms and vehicle safety testing.(2)Automatic driving scenarios construction method based on reconstructed accident scenarioStatistical analysis was performed on rear-end collision accidents,and20 typical rear-end collision accidents were screened out.Through the analysis of maps,trajectories,and vehicle dynamics parameters,the detailed motion state of the accident participants before the collision was deeply restored,and the position,orientation,speed,Acceleration,braking information,etc.,create a data foundation for the subsequent accident cause analysis and simulation scene construction.(3)Performance test of autonomous driving in high-risk scenarios in the virtual simulation environmentFrom the perspective of building autonomous driving test scenarios,the construction process of using Road Runner to design static road scenarios and using Open SCENARIO to edit dynamic traffic participant scenarios is introduced in detail.And the vehicle generation and dynamic trajectory algorithms are developed.The autonomous vehicle replaces the rear car in the simulation environment.The performance of the autonomous vehicle in this rear-end collision accident scenario is verified through the visual interface,driving record data,and radar perception data.Detect the state changes of surrounding vehicles to make correct decisions to avoid accidents.