基于驾驶数据的仿真场景自动化生成方法研究

虚拟仿真是智能网联汽车研发和测试的重要一环,仿真场景的数量多少和质量高低决定了自动驾驶策略的优劣。汽车产业有很多生成虚拟仿真场景的方法。本文依据《智能网联汽车数据格式与定义》(T/CAAMTB34-2021)团体标准进行了一种低成本、以数据驱动自动化生成场景的技术链路打通。该方案首先对自然驾驶中危险工况场景数据进行采集,并对其进行分析处理,依据标准生成结构化数据。该结构化数据可以批量自动化生成符合国际通用格式的OPENX系列标准的仿真场景,为后续大规模数据驱动测试场景自动化生成奠定了基础。

Virtual simulation is an important part of the development and testing of smart connected vehicles,and the number and quality of simulation scenarios determine the strengths and weaknesses of autonomous driving strategies.There are many methods for generating virtual simulation scenarios in the automotive industry.This article presents a lowcost,data-driven,automated scenario generation technology link based on the standard"ntelligent Connected Vehicle Data Format and Definition"(T/CAAMTB 34-2021).The solution first collects the scene data of dangerous working conditions in natural driving,and then analyzes and processes it to generate standard structured data.The structured data can be used to generate simulation scenarios in batch automation in accordance with the international common format of OPEN X series standard,which lays the foundation for the subsequent automated generation of large-scale data-driven test scenarios.