Thursday, April 18, 2024

AUKUS partners test swarms of AI-fueled drones in a ‘world first’ trial

The United States, Britain, and Australia have carried out the first AUKUS artificial intelligence (AI) and autonomy trial with the aim of rapidly driving these technologies into responsible military use.

During the trial, the AI-enabled assets operated together as a swarm to successfully detect and track military targets in a representative environment in real-time. Organized by the UK’s Defence Science and Technology Laboratory (Dstl), the trial achieved “world firsts,” including the live retraining of models in flight and the interchange of AI models between AUKUS nations.

The first-of-its-kind trial took place in April 2023, involving more than 70 military and civilian defense personnel and industry contractors. The trial utilized a variety of air and ground vehicles to test target identification capability.

The AUKUS Advanced Capabilities Pillar, known as Pillar 2, is pursuing a trilateral program of work on a range of leading-edge technologies and capabilities to promote security and stability in the Indo-Pacific region. Through Pillar 2, the three countries have collaborated to accelerate collective understanding of AI and autonomy technologies and how to rapidly field robust, trustworthy AI and autonomy in complex operations while adhering to the shared values of safe and responsible AI.

By sharing AI – and the underpinning data to enable it – with one another, Australian, UK, and US militaries can access the best AI, reduce duplication of effort, and ensure interoperability.

“The AUKUS AI and Autonomy trial in Salisbury Plains demonstrated AI algorithms working in a mission-tailored adaptive capability. The AUKUS research and operator teams collaborated to develop, test, and evaluate joint machine-learning models and operate our different national platforms on the battlefield,” said Hugh Jeffrey, Australian Deputy Secretary, Strategy, Policy, and Industry.

“I was impressed to see AI models rapidly updated at the tactical edge to incorporate new targets, which were immediately shared among the three partners to deliver decision advantage and meet changing mission requirements. This cooperation under AUKUS Pillar II will deliver a capability greater than any one country could achieve alone, and this really is the rationale for the AUKUS partnership at work.”

The trilateral teams collaborated to develop joint machine-learning (ML) models, apply test and evaluation processes, and fly on different national UAVs. The machine learning models were quickly updated to include new targets and shared among the coalition, and AI models were retrained to meet changing mission requirements.