The Research Of Simulation System For Autonomous Vehicle Based On UC-win/Road

With the rapid development of autonomous vehicle technology,the test requirements and the test tasks for autonomous vehicles increased dramatically.The industry generally believes that optimizing autonomous driving systems requires 11 billion miles of testing.It means that100 autonomous vehicles continuously test 500 years.Meanwhile,there is a problem that autonomous vehicles are difficult to cover the dangerous test scenarios when conducting road tests.Simulation for automated driving tests can reduce test costs,improve test efficiency,and increase the safety of autonomous vehicles.This thesis analyzes the typical structure of an automatic driving system and the composition of an automatic driving simulation test system.Then we propose an improvement plan for an automatic driving simulation system based on a domain controller test system,which use UC-win/Road software to build a virtual scenario.The main research works are described as follows:(1)In order to improve the accuracy of the vehicle motion model in UC-win/Road software and reflect the characteristics of the battery and motor in the vehicle,this thesis explores the method of modeling pure electric vehicles using Matlab/Simulink and Car Sim/Truck Sim software.We built model and validate performance for TX6790EV2 pure electric city bus.(2)To meet the needs of vision-based autonomous driving system simulation,this thesis uses Delphi 10.2 Tokyo development environment and UC-win / Road SDK tools to develop vehicle camera plug-ins for UC-win / Road software.The camera plug-in can set the camera rendering mode,basic parameters,installation position and angle in the vehicle,and use UDP packets for picture transmission.This article uses the camera simulation plug-in to deploy the YOLO V3 tiny model on NVIDIA Jetson Nano to realize real-time vehicle detection.(3)It has the demand for autonomous driving system to obtain V2 X information in the simulation process.This thesis uses Delphi 10.2 Tokyo development environment and UC-win/ Road SDK tools to develop V2 X plug-ins for UC-win / Road,which contains V2 V and V2 I in V2 X information.The V2 X plug-in can obtain the vehicle information between current vehicle and the intersection at the front road,and the traffic light information for the intersection.This system can be used to carry out the simulation test of the autopilot system based on vision and the research of CVIS.