jump to content

Agents@run.time: Trustworthy and Autonomous Artificial Intelligence in Digital Twins of Cyber-Physical Systems


Focus: Artificial Intelligence Talents
Type of funding: Individual funding programmes
Programme: CZS Research Boost
Funded institution:
  • Trier University of Applied Sciences

Prof. Dr. Thomas Hartmann of Trier University of Applied Sciences is developing methods for trustworthy and autonomous AI in cyber-physical systems (CPS). As part of this work, he is conducting research on knowledge-graph-based digital twins that can be used to monitor AI models at runtime and validate their decisions.

Goals

In the project “Agents@run.time,” Prof. Hartmann is developing the methodological foundations for combining the rigor, safety, and controllability of model-based systems with the capabilities of modern AI. The goal is to ensure trustworthy behavior, runtime monitoring, and sovereignty in AI-supported CPS models.

To this end, AI constraints are being investigated so that domain models can unambiguously map the decision boundaries of AI components and prevent hallucinations, lack of generalizability, and logical violations. Digital twins monitor AI behavior in real time and detect both drift and constraint violations. AI-enabled digital twins utilize edge inference, hybrid orchestration, and access controls to knowledge graphs to enable sovereign operation. Decisions can be traced back, even if full explainability is not achievable.

The architecture is evolving from models@run.time to agents@run.time, enabling a trustworthy, autonomous AI that is embedded in operational system models and accountable to them. The concept is being validated in the CPS laboratory at Trier University of Applied Sciences.
 

Involved persons:

Portrait
Portrait

Dr. Monika Schatz

Program Manager

Phone: +49 (0)711 - 162213 - 03

E-mail: monika.schatz@carl-zeiss-stiftung.de

Prof. Dr. Thomas Hartmann

Hochschule Trier

Detailed information:

Target group: Professors
Funding budget: 200.000 €
Additional overhead: 40.000 €
Period of time: March 2027 - February 2029

Funded institution: