Human-aware robust AI for automated driving
The advent of autonomous vehicles is one of the most anticipated technological developments of our time. But despite much progress in recent years, highly automated driving systems are still a long way from being able to drive freely on the streets of our cities. A major bottleneck on the road to the many potential benefits of these systems is their ability to share the road safely and efficiently with human-driven cars and pedestrians. This is a very complex issue, in which AI can play a decisive role.
HERALD Lab develops novel AI methods that will enable automated driving systems to interact responsibly and robustly with the people around them. To do this we will combine machine learning, cognitive modelling, control theory and multi-agent simulations. Ultimately, our goal is to lay the groundwork for a new generation of autonomous vehicles that are truly reliable partners for the people around them.
The HERALD Lab is part of the TU Delft AI Labs programme.
- Cognitive modeling of traffic interactions
Automated discovery of behavior prediction metrics for autonomous vehicles
- In addition to the projects above, we are always open to conversations with MSc students interested in doing a research assignment/graduation project aligned with our research directions
- Hybrid AI for pedestrian behavior prediction
- Modeling metacognition during decision making in traffic
- The role of attention in drivers’ decision making during merging
Merijn van Niekerk
- Investigating interactive merging behavior in a coupled driving simulator