Federated learning beyond healthcare: hydrogen fuel cells for commercial trucks

FedX had been applied exclusively in the healthcare sector. Expanding to automotive meant proving the platform's portability across verticals with fundamentally different data and engineering requirements. The specific target: identifying the optimal compression level that minimises hydrogen consumption for a given net energy output in commercial truck fuel cells - a performance optimisation requiring AI trained on real operational data from trucks in the field. That data lives at regional Tier-1 automotive partner's cloud infrastructure across multiple sites; centralising it raises IP, sovereignty, and data governance concerns that make the conventional approach unworkable.

E-Group and itsTier-1 automotive partner established a research-grade federated learning pilot under the IPCEI-CIS framework. The partner handles truck data collection, processing, filtering, and transfer to their regional cloud; the FedX client deploys regionally and trains models locally against this data. Only aggregated model updates are returned - raw operational data does not move. The system includes training and deployment of the federated model, monitoring dashboards for hydrogen consumption performance and model quality, and automated model retraining based on real-world performance feedback.

The pilot delivers value on two strategic fronts: it proves FedX's cross-vertical portability beyond healthcare into the automotive sector, broadening the platform's commercial relevance; and it positions E-Group as a contributor to the EU's clean energy and hydrogen transition through measurable efficiency gains in commercial vehicle fuel cells. Use Case and System Plan confirmed May 2026; working model with trained and deployed AI and full monitoring dashboard targeted August 2026. The project plans explicitly for its next phase: transition from this research pilot to industrial-scale deployment.