Digital Infrastructure and Decision Support

5G gives the low-latency, deterministic performance that closed-loop control needs on the factory floor, where jitter above 100 microseconds degrades quality. With 6G the wireless system itself becomes a sensor for position, direction and speed, which opens new ways to monitor and control.

The same infrastructure carries field data on product usage and health into maintenance and remanufacturing decisions, and enables over-the-air product updates.

AI surrogate models cut simulation times from hours to minutes. Where operational data is missing, for new products or rare failure modes, synthetic data from simulation and digital twins enables early exploration before production. Hybrid approaches combine physics-based models with data-driven methods where neither alone suffices. The track also covers the life cycle of the models themselves: versioning, validation, drift monitoring and retraining.

Cleaning, structuring and contextualising heterogeneous life-cycle data is underestimated work, and large volumes with little annotation are of limited use for learning. The track develops automated techniques such as transfer learning and domain adaptation for learning from scarce labelled data, together with the methods for data quality, traceability and explainability that AI-supported decisions require.

Decision authority stays with human stakeholders through transparent, traceable mechanisms. The track works on meaningful oversight, calibrated trust in AI recommendations and effective collaboration between people and AI, and on what future engineers must learn in order to use data-driven decision support well.