Scaleout Systems demonstrates autonomous AI for drone combat

Scaleout Systems, a NATO-backed startup founded by researchers from Uppsala University, is developing lightweight AI models designed to run directly on drone hardware. This technology allows unmanned aerial vehicles (UAVs) to perform reconnaissance and strike missions autonomously, even when disconnected from central servers or facing electronic jamming.

In a recent demonstration as part of the BAE Systems Bofors ‘Affordable Loitering Modular Ammunition’ (ALMA) program, a drone equipped with Scaleout’s AI successfully detected, identified, and ranked potential threats. The system autonomously selected an armored engineering vehicle as a high-priority target and executed a strike without real-time human intervention.

The system utilizes a federated learning architecture. This allows drones to process sensor data at the edge, performing local machine learning tasks. When communication is restored, updated models can be synchronized with central command nodes, which aggregate data to retrain and improve the models for future deployment. This approach aims to provide a sustainable combat advantage by allowing AI to adapt to changing environments, such as transitioning from desert to urban settings, within a single operation.