Anomaly Detection and Quality Indicators for Digital Maps Used in ADAS Applications

With the evolution of Advanced Driver Assistance Systems (ADAS), the gap towards Autonomous Driving (AD) is continuously narrowing. This progress is made possible using digital maps as one of the critical sources along with other ADAS sensors. Correct map data is crucial for the proper functioning of ADAS functions. This demands the need to evaluate the correctness of the map data regularly and efficiently. This work proposes a framework to quantify the map data correctness systematically. The framework algorithmically detects error locations in a map database and then derives KPIs from these error locations. The framework helps to identify issues in the map data related to the internal data consistency or heuristic rules. The framework consists of process automation in Python and map database checks in SQL. The proposed framework defines validation methodology that achieves goals like: (1) KPIs for map data reliability (2) systematic error identification. The framework was evaluated with maps from various sources. The framework yields results quickly and efficiently so that it can be regularly executed well before vehicle testing. In addition, the efficient KPI calculation permits the control of relevant map properties over subsequent map releases.