After three years of intensive collaboration, the project partners of the research project KIsSME (Artificial Intelligence for the selective real-time recording of scenario and maneuver data in the testing of highly automated vehicles) were able to celebrate its successful completion this week. At a closing event at Bosch in Renningen, the results of the joint project involving various partners from industry and research were presented.

The goal of KIsSME was to optimize data acquisition and processing during test drives of autonomous vehicles through the use of AI methods. By selectively filtering the sensor and environmental data for the truly safety-critical driving situations, the “data clutter” can be significantly reduced. This leads to efficiency improvements in the subsequent evaluation of the test drives.

Our company, among other contributions, has developed an AI algorithm that was continuously trained to be able to assess driving situations as reliably as possible based on their criticality. The quality of this criticality assessment is crucial to trigger the data recorder only when safety-critical maneuvers or system states are present.

“It was a pleasure to be part of this high-caliber consortium and to be able to contribute to the overall success,” explains Dr. Frank Hantschel, project manager at RA Consulting. “The results are very promising for future developments of systems for autonomous vehicles. Therefore, it is obvious that we will further deepen our collaboration with some partners in subsequent projects.”