High-definition maps are essential for the precise and reliable operation of Level 2+ automated driving systems. However, traditional HD map generation methods, which rely on survey vehicles, are expensive, time-consuming, and prone to rapid obsolescence as road conditions change. This presentation introduces two solutions developed by Bosch to address these challenges and enable scalable, real-time HD mapping. First, we present a crowdsourced mapping solution deployed across Europe, which leverages data from a fleet of user-operated vehicles to generate and update HD maps dynamically – eliminating the need for dedicated survey vehicles. Second, we discuss AI solutions that allow real-time onboard map generation using camera-based perception. This solution enhances accuracy and consistency by incorporating lightweight priors from existing SD maps or crowdsourced HD maps, allowing vehicles to adapt instantly to changing road conditions.

In this session, you will learn more about:

• Technical limitations and scalability challenges of traditional HD maps
• Data pipelines and processing strategies for generating and updating crowdsourced maps
• The benefits of combining crowdsourced maps with AI-based onboard map generation