Small AI Models Enable Drones to Identify and Strike Targets
▼ Summary
– Scaleout Systems, a NATO-backed startup founded by Uppsala University researchers, has pivoted to defense applications following Russia’s invasion of Ukraine.
– The company develops lean machine learning models designed to run on edge hardware in drones and forward bases for target detection and surveillance.
– Unlike competitors using large frontier AI models, Scaleout focuses on efficient computer vision tasks that fit within the constraints of small embedded devices.
– In 2025, Scaleout was selected for NATO’s DIANA Challenge Program to work on the Federated Aerial Intelligence for Recon project.
– CEO Andreas Hellander emphasizes the strategic advantage of operationalizing edge sensor data for NATO allies in modern warfare scenarios.
European defense capabilities are shifting toward edge AI integration, with a NATO-backed startup playing a pivotal role in equipping drones for autonomous target identification and engagement. Scaleout Systems is developing lightweight artificial intelligence models that allow small unmanned aerial vehicles to perform complex surveillance and strike missions without relying on cloud connectivity. This approach ensures that critical data processing happens directly on the device, reducing latency and vulnerability in contested environments.
The company’s origins lie in academia, having been established in 2018 by researchers from Uppsala University in Sweden. Initially, their work centered on training machine learning models to run on the onboard computers of commercial trucks and other ground vehicles. However, the geopolitical landscape changed dramatically following Russia’s full-scale invasion of Ukraine in 2022, prompting a strategic pivot toward military applications. The conflict highlighted the urgent need for robust, on-device intelligence that could operate independently of centralized networks.
Andreas Hellander, cofounder and CEO of Scaleout Systems, emphasized the strategic necessity of this shift during an interview with Ars Technica. “With the war in Ukraine and a shifting world, we realized that this technology can be very important to operationalize edge data and sensor data for machine learning to make sure that NATO allies have found that strategic advantage,” Hellander said.
Rather than depending on massive, resource-heavy frontier models from companies like OpenAI or Anthropic, Scaleout focuses on leaner AI architectures. These models are designed to execute computer vision tasks on the limited hardware available in drones or forward-deployed computing stations. According to Hellander, the systems must be adaptable to a wide range of infrastructure. “They need to fit on forward-deployed hardware and edge hardware, which can vary quite a bit from small embedded devices to quite powerful edge workstations,” he explained.
This versatility was recognized in 2025 when Scaleout was selected for NATO’s Defence Innovator Accelerator for the North Atlantic (DIANA) Challenge Program. Within this initiative, the company has been working on the Federated Aerial Intelligence for Recon project. The goal is to optimize machine learning algorithms for the specific constraints of drone processors, pilot tablets, and field command posts, ensuring seamless integration across various platforms used by allied forces.
(Source: Ars Technica)