DARPA-developed artificial intelligence has progressed from controlling simulated F-16s in virtual dogfights to piloting an actual F-16 in flight.
To continue reading the rest of this article, please log in.
Create free account to get unlimited news articles and more!
Air Combat Evolution (ACE) program developers uploaded AI software into a X-62A modified two-seat F-16 test aircraft at Edwards Air Force Base, California in December last year.
The AI was able to fly multiple flights over several days in the variable in-flight simulator test aircraft, providing live-flight data and demonstrating that AI agents can control a full-scale fighter jet.
ACE program manager Air Force Lieutenant Colonel Ryan ‘Hal’ Hefron said DARPA has been developing the program for the last three years since 2019, alongside broader testing with USAF Test Pilot School and the Air Force Research Laboratory.
“Thanks to the outstanding teamwork and coordination between DARPA, the Air Force Test Pilot School, the Air Force Research Laboratory and our performer teams, we’ve made rapid progress in phase two across all areas of the ACE program,” he said.
“VISTA allowed us to streamline the program by skipping the planned subscale phase and proceeding directly to a full-scale implementation, saving a year or more and providing performance feedback under real flight conditions.
“We conducted multiple sorties (take-offs and landings) with numerous test points performed on each sortie to test the algorithms under varying starting conditions, against various simulated adversaries, and with simulated weapons capabilities.
“We didn’t run into any major issues but did encounter some differences compared to simulation-based results, which is to be expected when transitioning from virtual to live.
“This highlights the importance of not only flight testing advanced autonomous capabilities but doing so on testbeds like VISTA, which allowed us to rapidly learn lessons and iterate at a much faster rate than with other air vehicles.
“Across all domains, research moves only as fast as the tools permit. VISTA’s recent upgrade makes it a much more effective test bed by enabling rapid integration and safe testing of AI-driven autonomy. It allowed us to accelerate full-scale flight test of AI-driven autonomy by at least a year.”
The X-62A aircraft can be programmed to demonstrate the flight-handling characteristics of a variety of different aircraft types and uses simulation programs to test autonomous F-16 AI agents. A safety pilot remains onboard to take control in an emergency.
As part of broader research, the US Test Pilot School is also measuring how much pilots trust the AI agent to conduct dogfighting or within-visual-range air combat while the human pilot focuses on larger battle management tasks.
US Air Force Test Pilot School director of research, Dr Malcolm Cotting, said TPS hosted an ACE Trust Capstone event last month using simulators to gauge pilot-agent alignment and additional trust-calibration flights in the X-62A are planned for later this year.
“The X-62A VISTA team has proven with this test campaign that they are capable of complex AI test missions that accelerate the development and testing of autonomy capabilities for the DOD,” Dr Cotting said.
“The X-62 is a flight test accelerator, it allows the USAF to rapidly execute tactical autonomy algorithm and fixed-wing vehicle model flight tests.
“These tests generate data that is used to improve the algorithms and vehicle models at a remarkably fast pace.
“The X-62 illustrates the value of using crewed, surrogate vehicles that possess tactical performance characteristics as autonomy test assets.”
The ACE program ran AlphaDogfight Trials in August 2020 where AI agents flew simulated F-16s in a virtual dogfighting competition. The finale of the trials was an AI agent engaging and defeating an experienced F-16 fighter pilot flying in a simulator (5-0).