In NEUPA, short for "new explainability supported by world models for predictions in autonomous driving", we develop methods to increase explainability in AI systems.

Based on Methods from Machine Learning, we would like to predict dangerous behaviors of pedestrians, cyclists, and e-scooter riders for autonomous driving. We also investigate methods from statistical physics and collective decision making to analyze and explain these methods (e.g., Artificial Neural Networks) to make them certifiable for safety-critical tasks. More details can be found on NEUPA.net.

Team & Contact

• Prof. Berekovic
• Prof. Hamann
• Dr.-Ing. Mostafa Wahby (wahby@iti.uni-luebeck.de)
• Ulrich Behrje
• Julian Petzold