Automated vehicles have the potential to fundamentally change our traffic. Proving safety is necessary for the acceptance and use of such systems, but involves major challenges. Industry and research are focusing on scenario-based testing, in which a catalog of relevant traffic scenarios is built up. In this environment, we conduct research on topics related to the construction, efficient use, and simulation of the scenarios.
Scenario extraction from real data
Which traffic scenarios must vehicles be able to complete without accidents? Frequencies, relevance and criticality of such scenarios can be captured by analyzing recorded data. For this purpose, we research on grouping similar scenarios by clustering and on finding unknown scenarios by anomaly detection.
Behavior description maneuver
The description of the behavior of road users is an elementary component for the development and testing of automated driving functions. We are researching a form of description that can be used throughout the entire development process by abstracting the behavior to maneuvers. Possible applications are e.g. the comparison and simulation of traffic scenarios.