Mixture of Experts models Tailored for Fleet Intelligence

Purpose and goal

Digitalization is generating massive amounts of data at the network edge. Federated Learning (FL) enables local AI training without sharing raw data, enhancing privacy across sectors like healthcare, transport, finance, and defense. Scaleout Systems is advancing fleet intelligence, where vehicles collaborate for safety and monitoring. By combining FL with Mixture of Experts (MoE), the project builds scalable, efficient, and privacy-focused edge AI solutions.

Expected effects and result

This work has the potential to drive fundamental change in federated learning. By advancing Mixture of Experts (MoE) within FL, the project will position Scaleout and Sweden at the forefront of next-generation federated AI solutions. It will redefine how distributed intelligence is developed, scaled, and deployed globally across sectors requiring privacy and efficiency.

Planned approach and implementation

The project will be carried out through well-defined work packages (WPs), each addressing specific technical and operational goals towards a novel federated MoE-based perception architecture for fleet intelligence. Collaboration with AI Sweden and Zenseact as use-case partner will ensure technical strength and real-world relevance. Regular joint workshops and meetings will support knowledge transfer, innovation, and project alignment.