Construction Of Simulation Test Scenario And Risk Assessment Of Autonomous Vehicle

At present,the incidence of traffic accidents is high and traffic congestion is serious.The development of self-driving cars will effectively solve the above problems.The development of self-driving cars is a gradual process,and its safety on the road is always a concern.However,the complex traffic environment brings great challenges to the safety of autonomous driving.How to test the safety of self-driving cars in different driving scenarios is a key issue in the development of self-driving cars before they are fully on the road.At present,the test of self-driving cars is mainly divided into virtual simulation test,site test and public road test.Currently,the test scenario library is not complete,and there is also a lack of systematic research on the assessment of scenario risk.This paper takes the risk test scenarios of self-driving cars as the research object.Firstly,it analyzes the risks of self-driving cars in the process of driving,and studies the driving risks of self-driving cars in typical road scenarios,severe weather scenarios and emergency scenarios.Second in automatic car driving test scenario elements on the basis of the description and layered,stratified sampling method is adopted to develop new test scenario,consider a typical traffic characteristics of static and dynamic test scenarios,city road,highway,mountain roads and dynamic scene(free driving,following and lane changing,etc.)is the design of the scheme.Again,the paper introduces the automatic car driving simulation structures,key technology,the virtual test scenarios using SCANe Rstudio software based on virtual simulation platform for the development of related scenario and automatic car driving simulation experiment,the simulation results are analyzed.Finally,based on the simulation results,fuzzy comprehensive evaluation is used to further judge the risk and importance of the scene.The risk evaluation index and method are proposed for the dynamic scene with high risk degree after the judgment,and the specific risk index calculation and risk evaluation are carried out for the self-driving car following driving scene based on the simulation data.The effectiveness of the risk evaluation is determined by comparing with the actual simulation results.The research results show that:when the TTC threshold of the evaluation index based on time is 4s and the calculated value of TTC is greater than 4s,the driving risk of the autonomous vehicle is low;when the calculated value of DRAC based on deceleration is 8.45m/s~2 and the calculated value of DRAC is less than 8.45m/s~2,the driving risk of the autonomous vehicle is low.