This article is more than 1 year old
Wow, so they actually let AI fly an F-16 fighter jet
You can be my Bing man anytime
Pentagon boffins have for the first time used AI algorithms to automatically control a real F-16 fighter jet mid-flight. Well, OK, at least for the first time they can talk about.
The aircraft flown in the experiment, dubbed X-62A or VISTA, was modified and equipped with the right hardware components to run the software developed under the Air Combat Evolution (ACE) program run by the US Dept of Defense's science nerve-center DARPA.
Launched in 2019, the program's goal is to revamp aircraft combat by enabling F-16s to automatically dogfight. DARPA envisions machine-learning algorithms assisting pilots with flying and perform tactical maneuvers, with our humans focusing on battle commands, strategy, and firing weapons. This work involves developing new software and models.
Last year, an AI agent beat a real US Air Force instructor in a virtual dogfight conducted in flight simulation. And in December, DARPA managed to use this new technology to help fly a real F-16 jet, closing the simulation to real-world gap.
Air Force Lt. Col. Ryan "Hal" Hefron, the DARPA program manager for ACE, said the AI algorithms were used during takeoff and landing to fly the VISTA aircraft. So, like a fancy autopilot, perhaps? Multiple test flights were conducted over several days at the Air Force Test Pilot School (TPS) at Edwards Air Force Base in California last December.
"We conducted multiple sorties [takeoffs 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," Hefron said in a statement.
"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."
We're taking it as read here that DARPA's AI algorithms are a class beyond the traditional, non-intelligent routines used in today's aircraft autopilot.
- Take my bits awaaaay: DARPA wants to develop AI fighter program to augment human pilots
- NASA's latest AI will navigate the Moon using landmarks
- US warns aging air-traffic control code won't be fixed until 2030
- US Air Force reveals B-21 Raider stealth bomber that'll fly the unfriendly skies
DARPA worked with scientists and engineers from academia and industry, including EpiSci, PhysicsAI, Shield AI, and the Johns Hopkins Applied Physics Laboratory to test different types of AI algorithms. The modified two-seater VISTA jet comes equipped with the System for Autonomous Control of Simulation (SACS), the computer system used to run the software. A pilot was onboard at all times to monitor the software's performance and take over if needed.
DARPA is also testing how human pilots interact with AI to examine how well they trust machines to automatically conduct dogfights. Air Force pilots have flown L-29 jet trainers also running AI algorithms at the University of Iowa Technology Institute's Operator Performance Laboratory.
The aircraft used in the experiments contain sensors in the cockpit to track the pilot's physiological responses as they fly. DARPA believes it will be able to figure out in what scenarios do pilots trust or not trust AI by monitoring their flight actions and the sensor data. The agency is going to conduct similar experiments on test pilots flying the VISTA aircraft later this year.
"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 2 across all areas of the ACE program," said Hefron.
"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."
The Register has asked DARPA for further comment. ®