FREEPORT: Federated Learning and Edge Processing for Safe and Efficient Operation

Important results from the project

The project successfully met all proposed objectives, advancing edge analytics and federated learning for heavy-duty vehicles. Edge data collection systems were deployed on test and customer vehicles. An online anomaly detection algorithm was validated through a live demonstration, showcasing real-time monitoring of energy consumption in a mining environment. Edge Processing was demonstrated on road slipperiness monitoring using ten MX4 edge devices during the AI Sweden event in Gothenburg.

Expected long term effects

The project’s results are expected to generate industrial impact by developing the edge analytics and federated learning frameworks that enable scalable, privacy-preserving AI in commercial vehicles, improving efficiency, safety, and sustainability. The project improved predictive maintenance and energy monitoring, which reduces downtime and emissions while extending asset lifespan. These outcomes lay the foundation for intelligent, resilient, and sustainable systems for edge analytics.

Approach and implementation

The project was implemented in close collaboration between Volvo Group Truck Technology, Volvo Trucks, Boliden AB, Halmstad University, RISE, Stream Analyse, and Lindholmen Science Park (AI Sweden). The work covered both technical implementation, such as edge data collection and processing, and methodological development of advanced AI models, ensuring a secure and efficient framework for real-world deployment.

External links

Federated Learning & Edge Processing for Safe and Efficient Operation