Project Safe AI Engineering
Safe AI Engineering is part of the AI family. It was initiated and developed from the VDA's flagship autonomous and connected driving initiative. The project is funded by the Federal Ministry for Economic Affairs and Energy. Twenty-four scientific and business partners are developing a comprehensive methodology to safeguard the entire life cycle of AI functions. A key area of innovation is the standardisation of training and validation data to enable more sustainable and cost-effective usage. The methods developed will be tested and validated using an AI perceptual function for recognising pedestrians as an example.
Further information can be found at https://safe-ai-engineering.de.
Contribution DLR-Institute for AI Safety and Security
In addition to coordinating with Luxoft, the Institute for AI Safety and Security plays a central role in orchestrating the project. It is actively involved in further developing the fundamental AI engineering methodology, defining use cases, and implementing them throughout the entire lifecycle of an AI component.
A key focus lies on normalizing data and evaluating the associated information quality, especially regarding domain adaptation and transformation. Additionally, the team is working on methods for AI monitoring—particularly for tracking explainable uncertainty estimates—and conducting risk assessments based on uncertainty quantification.
Institutes and facilities involved (DLR & external)
- German Aerospace Centre, Institute for AI Safety and Security
- Akkodis Germany GmbH
- AUMOVIO SE
- AVL Deutschland GmbH
- Bundesanstalt für Straßen- und Verkehrswesen
- Bertrandt Ing.-Büro GmbH
- Robert Bosch GmbH
- Capgemini Engineering Deutschland S.A.S. & Co KG
- Cariad SE
- DXC Luxoft GmbH
- Fraunhofer-Gesellschaft e.V. (IESE und IKS)
- FZI Forschungszentrum Informatik
- Karlsruhe Institute of Technology (KIT)
- Mercedes-Benz AG
- Opel Automobile GmbH
- Porsche AG
- Spleenlab GmbH
- Technische Universität Berlin
- Technische Universität Braunschweig
- TÜV AI.Lab GmbH
- Valeo Schalter und Sensoren GmbH
- ZF Friedrichshafen AG
(c) BMWE