面向高精地图与融合定位的测试场景库研究

打开文本图片集

中图分类号： 文献标识码： 文章编号： （ ）

Research on Test Scenario Database for High-definition Mapsand Integrated Localization

Cai Cong，Zhang Dengquan，Wang Qingshan，Xia Haipeng，Wu Jiayi （CATARC（Tianjin）Automotive Engineering Research Institute Co.，Ltd.，Tianjin 300300，China）

【Abstract】 This study addresss the testing requirements for high-definition （HD）maps and integrated positioning technologies byproposing a comprehensivescenario databaseconstruction methodology tosupport multidimensional performance evaluation and optimization.By analyzing the interdependencies among localization，perception， and mapping technologies，the scenario database is categorized into localization scenarios，perception scenarios，and mapping scenarios，with parameterized designs based on key influencing factors such as weather conditions，signal occlusio，and dynamicenvironments.A bidirectional scenarioanalysis framework is introduced，enabling forward prediction of system performance from predefined conditions and backward inferenceof latent environmentalfactors from empirical data，therebysystematically validating adaptabilityincomplexenvironments.Additionally，progressiveand reinforcement learning-drivenscenario sequencing strategies aredeveloped tosimulatereal-world dynamic interactions among multi-module systems.

【Key words】localization；perception；map

0 引言

随着自动驾驶技术的快速发展，高精地图与融合定位技术成为实现高精度导航和定位的关键手段。（剩余3947字）