A Study On Intelligent Vehicle Algorithm Simulation Test Platform Construction And Algorithm Test In Accident Scene

With the increasing number of vehicles equipped with advanced assisted driving functions,the number of accidents caused by the reliability of the assisted driving system is also increasing.As the speed of research and development and the speed of technological evolution continue to accelerate,the vehicle assisted driving system is also moving toward with the development of complexity and diversification,traditional simple road scenarios and testing techniques have been unable to quickly and effectively test vehicles equipped with intelligent driving technology.Therefore,how to evaluate the intelligent vehicle control algorithm efficiently and comprehensively becomes an urgent problem to be solved in the development process of smart cars.This paper mainly expounds a smart car simulation test platform system based on model-in-the-loop theory,and introduces the vehicle model construction process,virtual test scenario construction,algorithm test and optimization process.In addition to the construction of the smart car simulation test platform,this article will also use the actual accident case to carry out accident recovery in Prescan,and replace the original accident vehicle with the test vehicle controlled by Apollo intelligent algorithm,through the whole model-in-the-loop real-time simulation system.To test the specific performance of the Apollo smart car control algorithm in the accident situation.Firstly,the research status of the test methods of smart cars and smart cars at home and abroad is investigated in detail,and the shortcomings and problems of the traditional methods are discussed.Then,the software-in-the-loop method is used to construct the whole set of smart car test simulation platform.Then the test platform construction ideas are introduced in detail: using Trucksim and Prescan software to build virtual test scenarios and vehicle models,Apollo open source platform construction and module selection,ROS communication system to achieve Linux system and windows system communication,Thereby the control algorithm controls the vehicle.Finally,according to the original data of the accident in PC-CRASH,the accident reconstruction is carried out and the scene GIDAS format data is exported.The GIDAS data is imported into the Prescan to reconstruct the accident scene,and then the original accident vehicle is replaced with the virtual vehicle model controlled by the Apollo control algorithm.The built model is tested in real-time in the ring platform,and then the Apollo control algorithm is optimized and modified according to the specific performance of the vehicle in the visualization result.Finally,the set of smart car model-in-the-loop test simulation system is tested in specific experiments.The performance is compared and discussed with traditional test methods.It proves that the test cycle of the smart car test simulation platform has a shorter test period while ensuring a certain precision.It is more suitable for rapid proof-of-concept and pre-function test of the whole vehicle algorithm in the early stage of smart car development,and the accident scenario simulation test.Joining has an important role in promoting smart car development.