We congratulate our Institute’s Thomas Monninger for his outstanding success in his doctoral dissertation: "Scene Understanding of Static Road Infrastructure for Autonomous Driving.", in a partnership between the Analytic Computing group of the Institute of AI of the University of Stuttgart and Mercedes-Benz.
Thomas’s work addresses a major problem in self-driving car development: the use of pre-recorded High-Definition maps, which can quickly become outdated due to events like construction. To solve this, Thomas suggested a method that creates maps on the spot using live sensor data, with help from offline maps that can be easily updated. By treating differences between old maps and live data as random changes in a smoothing process, his system lets self-driving cars travel safely even when maps do not exactly match the current roads.
The resulting framework induces a clear cycle: improved online perception leads to the development of accurate offline maps, which then provide strong guidance for the online system. This method not only enhances map quality but also offers an important measure of spatial uncertainty.
We wish Thomas all the best in his next scientific chapter as he continues to shape the future of autonomous mobility.
