Autonomous strike drone uses Nvidia Jetson Orin Nano to ...

Swedish artificial intelligence startup Scaleout Systems demonstrated an autonomous loitering munition that uses onboard computer vision to identify and prioritize targets. Developed in partnership with BAE Systems Bofors as part of the Affordable Loitering Modular Ammunition program, the strike drone relies on a small artificial intelligence model running on an Nvidia Jetson Orin Nano hardware board.

During recent tests, the drone used onboard processing to scan the environment, detect potential threats, and rank an armored engineering vehicle as the highest priority. Although the system can operate with minimal manual input, the demonstrations maintained a human in the loop. A human operator verified the chosen target and authorized the final strike approach before the drone navigated to the target's last known position to simulate an airburst munition deployment.

The project utilizes federated learning through a tactical computer vision network, allowing devices to train artificial intelligence locally and share model updates for rapid adaptation. Tests also demonstrated the system functioning successfully under extreme cold weather conditions and maintaining operations when network connections were degraded or completely cut.

These technological demonstrations occur ahead of any official NATO procurement order or actual combat deployment, raising broader discussions about autonomous weapons and the legal requirement for human oversight.