Washington: The US Air Force is planning to acquire around 1,000 AI-controlled armed drones in the coming years. These drones, known as “loyal wingman” drones, are expected to be more agile and cost-effective compared to traditional manned fighter jets. A recent experiment showcased the potential of AI in aerial combat.
In 2019, the US Defense Advanced Research Projects Agency (DARPA) initiated the
AlphaDogfight project, inviting eight teams of coders, including big defence firms like Lockheed Martin and Boeing, as well as smaller companies and university groups. Their task was to develop an AI specifically designed for dogfighting.
During mid-2020, DARPA organised a series of dogfighting competitions, where different AIs competed against each other. Subsequently, the most competitive AI, developed by Heron Systems based in Maryland, was pitted against an actual human pilot known as “Banger,” flying a US Air Force F-16.
The Heron drone, using aggressive tactics, swiftly targeted and shot down Banger. Unlike earlier drone AIs that struggled with unpredictability, the Heron AI showcased superiority by using direct attack techniques, which negated the human pilot’s advantage of creative thinking.
Besides, the XQ-58, a jet-propelled drone measuring 29-feet-long, serves as the primary test vehicle for the US Air Force’s own AI experiment called Skyborg, which has been ongoing for several years.
The goal of Skyborg is to develop cost-effective drones that can fly alongside manned fighters, enhancing their capabilities in complex and risky aerial battles. These drones are expendable, allowing for more aggressive manoeuvre and the ability to penetrate heavily defended airspace without risking human lives. The Skyborg programme is progressing rapidly and is expected soon to be adopted formally.