We are pleased to announce that, under the supervision of Bernd Ludwig, our Human-Centred Artificial Intelligence (HCAI) Master's student Noah Meißner (externer Link, öffnet neues Fenster) and Computer Science Bachelor's student Tim Sieber had a paper accepted and published at the 1st International Conference on Geospatial Artificial Intelligence (GeoAI 2026) (externer Link, öffnet neues Fenster).
Noah presented the research at the conference in Ghent, Belgium. Here is Noah's report from the conference:
"Pedestrian navigation works well—until you go indoors.
Last week, I presented our work at the first edition of the GeoAI 2026 conference in Ghent. It was a great start, with a strong mix of researchers and ideas.
Our paper: "Seeing Around the Corner — Fusing Visual Flow and Inertial Sensors for Indoor Pedestrian Navigation"
The core challenge: indoor environments break the assumptions that make outdoor navigation reliable. No GPS. Complex layouts. Sensor drift that compounds over distance.
What we built at the University of Regensburg:
→ A scalable pipeline that aligns HoloLens 2 positioning with smartphone sensor data
→ High-quality ground truth for training and evaluating indoor navigation models
→ ~90% accuracy in predicting turn-taking within defined turning zones
This is a step toward navigation systems that stay reliable over longer distances, in the environments where people actually need guidance: hospitals, airports, large buildings.
📄 Read the paper: https://zenodo.org/records/20313731 (externer Link, öffnet neues Fenster)
Big thanks to my co-authors Tim Sieber and Bernd Ludwig, and to the GeoAI organizers for putting together such a strong first edition."
Ghent University & Noah Meißner