The U.S. Air Force has demonstrated the application of artificial intelligence in aerial combat during the launch of an XQ-58A Valkyrie.
This flight marked a milestone in the development of a tactical autonomy algorithm, transitioning from theoretical simulations to practical testing on an unmanned aircraft.
The AI algorithms employed in this experiment were integrated into the XQ-58A and deployed in a live test at the Eglin Gulf Test & Training Range. Harnessing the power of deep reinforcement learning, these AI algorithms leveraged neural networks to autonomously pilot the unmanned aircraft in real-time combat scenarios, engaging simulated adversaries with mission systems and weaponry.
Ryan Bowers, the principal test engineer leading this endeavor, emphasized the imperative of amalgamating AI testing methodologies with conventional evaluation techniques, underscoring the wealth of insights gained during this initiative that will serve as guiding principles for future projects.
This flight test, executed by the 40th Flight Test Squadron with substantial support from AFRL and Kratos Unmanned Aerial Systems, built upon the success of a previous test flight on July 25. The earlier flight exhibited an AI-enabled, high-performance unmanned aircraft for the Department of Defense, adeptly showcasing standard aviation and navigation tasks while meticulously implementing safety protocols to prudently manage risks.