To increase safety in automated driving, TomTom and HELLA Aglaia teamed up to enable robust camera-based localization and crowdsourced HD map updates.
Customer story
Customer story
For safe automated driving, you need detailed maps that constantly reflect the reality on the road. One of the biggest challenges in HD mapping is therefore keeping the HD map up to date in real time. The time between a change in reality on the road, such as a new lane divider or speed limit sign, and when the map is updated is quantified as ‘reality to map’, or R2M. The shorter the R2M, the more accurate and therefore safer an HD map will be for automated driving systems.
At TomTom, we minimize R2M time in our HD Map by taking a multi-source approach to mapmaking and map maintenance, which includes crowdsourced car sensor data. To further enhance R2M time, we teamed up with HELLA Aglaia and combined their localization with our crowdsourced data for faster, highly accurate HD map updates.
How did we do it? First, we delivered the TomTom HD Map to a vehicle via TomTom AutoStream. HELLA Aglaia’s camera achieved perception using machine learning for detection and classification of lane dividers and traffic signs. Then, centimeter-level localization was completed by correlating the lane and sign observations with the TomTom HD Map. Lastly, real-world change detections in the form of Roadagrams were sent to the TomTom cloud.
Created by TomTom, Roadagrams are accurate crowdsourced map snippets from car sensors. The Roadagrams are aligned and aggregated, and then compared with our HD Map. Detected changes in the map are processed and added as new map attributes, such as new traffic signs. And finally, TomTom AutoStream delivers the HD Map updates back to the vehicle.
In short: Faster, more accurate HD map updates for safer automated driving. Such a result ensures that drivers’ maps for automated driving best match the reality on the road, being able to take the latest road changes into account while controlling a vehicle.