Scaleout Systems, a NATO-backed startup founded by Uppsala University researchers, has developed specialized machine learning models designed for autonomous target detection and selection on edge hardware. Moving away from massive frontier models from providers like OpenAI, the company utilizes leaner computer vision models capable of running directly on resource-constrained devices, including small drones, pilot tablets, and field command workstations. As part of NATO’s Defence Innovator Accelerator for the North Atlantic (DIANA) Challenge Program in 2025, the startup progressed the Federated Aerial Intelligence for Recon project. This technical approach enables drones to perform real-time battlefield surveillance and target identification locally. A critical component of their architecture involves the ability to transmit selective intelligence updates to local platoon or company headquarters rather than transmitting sensitive, raw sensor data. This methodology optimizes data usage and enhances security by operationalizing edge computing to process sensor information on-site. Originally focused on vehicle-based machine learning, Scaleout Systems pivoted to these defense applications following the 2022 invasion of Ukraine to provide NATO allies with a strategic advantage through decentralized, autonomous aerial intelligence.
Read original
arstechnica/ai