Research And Implementation Of Digital Twin Scene Interaction For Automatic Driving Simulation Test

Before a self-driving vehicle is put into commercial use,a large number of comprehensive tests including safety,system and function are required.The test for self-driving cars is also a popular research topic in both academic and industrial.In the driving test,safety is a particularly important issue.About 90% of the autonomous driving test algorithm needs to be completed on the simulation platform.The closed test field accounts for about 9%,and only1% is carried out in the actual road test.Therefore,the virtual test of autonomous driving plays an unreplaceable role for autonomous driving technology.At present,the autonomous functions that can be realized by vehicles are constantly increasing,and the level of autonomous driving technology is gradually improving,which bring more new test requirements.The traditional autonomous driving simulation test methods cannot meet the new test requirements due to scattered scenes and low reliability.In recent years,digital twin technology has emerged as a new technology hotspot for innovation and development in various fields,showing a very broad development prospect.Digital twin technology has key features such as virtual-to-real mapping,full life cycle,strong repeatability,and spatial visualization.In order to make up for the shortcomings of actual road tests,on the premise of ensuring safety,the measured traffic scenes are mapped to virtual space through digital twin technology,save testing time and costs,and promote the development of autonomous driving technology.This thesis proposes a test method for the current test methods of autonomous driving simulation test that cannot simultaneously incorporate the driver,vehicle dynamics,variable weather,and road into the test environment,or ignore the test cost and poor operability.Real vehicles,roads and road testing facilities are used as prototypes.Digital twin technology is used to construct prototype mapping objects（digital twins）with extremely high resolution in virtual space.The traffic flow constitutes a complex driving scenario,which is used as a test environment for autonomous driving.In the digital twin scenario for autonomous driving testing,network and communication technologies are integrated to realize the interaction between the elements of the digital twin scenario,including the following aspects:1.A digital twin road environment for autonomous vehicle testing is constructed,including the process of digital 3D modeling,collision detection and spatial coordinate transformation of the digital twin prototype vehicle in virtual space.Through the joint debugging of r Fpro and Car Maker,a road environment consistent with the real road network structure and latitude and longitude information can be constructed,and a complex autonomous driving test environment can be formed by setting parameters such as weather,road surface diseases,and traffic flow density.2.A communication link established for the digital twin scene elements in the autonomous driving test environment,which builds up an autonomous driving hardware-in-the-loop test system composed of Car Maker,r Fpro,and electronic control unit based on the Matlab/Simulink platform.After analysis and discussion.The UDP communication protocol is applied to complete the interaction between various modules in the test environment.3.Based on the communication link of the scene,the test of vehicle following and AEB vehicle hitting pedestrians was carried out.The experiment verified that the UDP communication protocol can well eliminate the delay generated in the communication and network transmission process.The test results show that the electronic control unit with the automatic driving algorithm in the device has high accuracy and error control ability.4.The digital twin scene interaction management system is implemented.Based on the architectural idea of separating the front and back ends,the front end is developed using the React technology stack,providing five visual functional interfaces for user management,scene management,test case,and test result presentation.The back end uses the Mongo DB database to store user information and test results of a large number of autonomous driving cases,and assign corresponding permissions by judging user roles and displays different visualization function pages.