Über Kleinanzeigen auf der Jagd nach neuen Minds? NEIN! Das sind nicht wir und bitte reagiert nicht darauf. Das ist fake! Wir suchen euch über die üblichen bekannten Plattformen, oder auch direkt in unseren Social Media Kanälen wie LinkedIn, Facebook, XING und Twitter. Findet uns dort und tretet in Kontakt mit uns. Wir freuen uns auf euch!

Completion of the KI Absicherung project – safety arguments for autonomous driving in Berlin

Completion of the KI Absicherung project – safety arguments for autonomous driving in Berlin

Within the “AI family”, the KI Absicherung project crossed the finish line on June 23, 2022, as the first of four projects. Embedded in the VDA flagship initiative for autonomous and connected driving and funded by the German Federal Ministry of Economics and Climate Protection, project participants from both the industry and the science sectors investigated key questions leading to securing AI-based functions for highly automated driving.

How does it work?
Deep Neural Networks play a decisive role in introducing highly automated and autonomous driving. They use various sensor data to understand and perceive and make decisions about the driving environment, including pedestrians or obstacles. This is the only way to make highly automated driving possible since the vehicle can and must react appropriately to different scenarios.

Why KI Absicherung?
The goal of the project was to achieve an industry-wide consensus on systematic methodology for the acceptance of AI functions in the field of autonomous driving. QualityMinds has been actively involved in the project for three years and has taken on various tasks within the “Synthetic Data” subproject. In addition to helping to set up a synthetic data set through requirements engineering and establishing a “quality gate”, we examined the so-called corner cases. These are scenarios in which the neural network reacts to its environment in an unexpected way. Our task was to identify cases and certain areas of the AI functions where the corresponding neural network does not work as expected and further action must be taken.

Thanks to this groundwork and the data set jointly created in the project important challenges to the realization of autonomous driving were identified. The validation methodology, that has now been publicly presented, together with the evidence for the derivation of systematic test and training methods, is now not only incorporated into national development, but it is also applied in international standardization, such as the ISO/PAS 8800.

Scientific publications stemming from the project were also published. To name one, Niels Heller and Namrata Gurung von QualityMinds presented their paper on the automatic creation of a corner case data set during the DATA 2022 conference in Lisbon.

QualityMinds would like to thank all 24 partners involved, as well as other project participants, coordinators, and sponsors. We will continue to conduct research in the AI field. One of our current projects includes “Attention – Artificial Intelligence for real-time injury prediction” where we work together with the Fraunhofer IAIS institute.

You can find more information at https://www.ki-absicherung-projekt.de/en/ or you can write us about it directly. We’re looking forward to your questions!

0 Comments

Submit a Comment

Your email address will not be published.

Autonomous driving

written by

Bastian Karl Knerr