ITSC 2025

The MRT recently participated in the ITSC in Gold Coast, Australia, presenting two papers on robust traffic-light detection using HD maps and on prediction-driven motion planning:

• Robust Traffic Light Detection and Stopline Release by HD-Map Association, Richard Fehler, Kevin Rösch, Fabian Immel, Christoph Stiller:

We developed a traffic light perception system for automated vehicles that combines deep learning and a Min-Cost-Flow matching algorithm for robust stop line release. Tested in real-world autonomous driving across two German cities, including challenging conditions like rain and low light, the system achieved 96.68% accuracy.

• Prediction-Driven Motion Planning: Route Integration Strategies in Attention-Based Prediction Models, Marlon Steiner, Royden Wagner, Ömer Sahin Tas, Christoph Stiller:

Combining motion prediction and motion planning offers a promising framework for enhancing interactions between automated vehicles and other traffic participants. This paper investigates the extension of attention-based motion prediction models with navigation information and bridges the gap between multi-agent motion prediction and goal-based motion planning. We propose and evaluate several architectural navigation integration strategies to our model on the nuPlan dataset.