The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Publications Copernicus
Articles | Volume XLIII-B4-2022
02 Jun 2022
 | 02 Jun 2022


J. M. Lógó and A. Barsi

Keywords: autonomous driving, HD map, topology analysis

Abstract. Autonomous and highly automated driving has become one of the key research topics – even in mapping sciences. Today, it has been clearly proven that the desired autonomy can be reached solely by progressive development, where maps and requirements against environmental models shall be modified. The research of the past decades resulted in the specifications for building high-definition (HD) maps, which contain all necessary field objects with their relevant features in a sophistically designed database. Maps have also become indispensable tools in reliable automotive development processes as simulations request accurate and detailed environment descriptions. The most accepted simulator map format is ASAM’s OpenDRIVE, having a complete ecosystem nowadays with ultra-fine resolution pavement surface model, traffic flow description, and essential modules to produce those components by mainly automatic data collection and processing. Simulations enable efficient analysis of vehicle behavior as well as testing workflow for (onboard) vehicular and infrastructure sensors. With respect to the requirements of HD maps, not only their geometric fidelity but also the correct topology is expected. The available technology already serves OpenDRIVE models for various scenarios, where the topologic correctness is hard to test. The current research puts the emphasis on the topology analysis: with parsing the existing models, topology descriptors are derived, then a test suit containing rules of acceptable cases is applied. The rule set is built of items for detecting errors and warnings considering topology parameter tolerances; the result is quality documentation after a comprehensive testing approach. The actual topology testing environment forms an excellent base for (semi)automatic error fixing platforms involving artificial intelligence.