Trustworthy Autonomous Systems

With this in mind, the Trustworthy Autonomous Systems field of competence develops trustworthy, autonomous/cognitive software architectures on the basis of artificial intelligence (AI). We conduct research into innovative approaches for the planning of intelligent behaviors and develop scalable, dependable algorithms for the merging of sensor, data and information. Our researchers create structured approaches that combine state-of-the-art system engineering with the latest AI research and supply the reference for development and certification.

We also conduct research into implementation concepts for autonomous/cognitive behavior, in addition to verification methods for autonomous/cognitive systems, which are linked at three levels:

• solution: behavior-dynamic simulation approaches that use intentions as the ground truth
• system: architecture approaches and redundancies
• technology: methods related to error models from systematic, chance and uncertainty

We test our approaches on application demonstrators such as fully-automated driving and autonomous flying, as well as in intelligent infrastructures and digital twins.