DART: Distribuerad AI för Robusta och Tillförlitliga autonoma fordon

Important results from the project

The project achieved its goal of establishing a strong foundation for Swedish-American collaboration on distributed AI in autonomous systems. In collaboration with NVIDIA FLARE, a concept for privacy-preserving and cross-silo model training was defined and validated. The project initiated international collaboration and provided valuable insights into scalable and trustworthy AI, supporting future development and a continued Stage-2 initiative.

Expected long term effects

The project is expected to enable long-term adoption of distributed, privacy-preserving AI across multiple industries. By leveraging NVIDIA FLARE, it supports scalable and trustworthy model development without the need for data sharing or centralization. The results contribute to more robust AI systems, strengthen international collaboration, and position Swedish actors for future innovation.

Approach and implementation

The project was implemented as planned in three phases: preparation, on-site collaboration with NVIDIA FLARE, and follow-up. Activities followed the defined structure and the schedule was maintained without major deviations. The collaboration worked well, enabling effective knowledge exchange and technical alignment. No major external factors affected the project, and minor adjustments were managed without impacting overall progress.