• Thanks for stopping by. Logging in to a registered account will remove all generic ads. Please reach out with any questions or concerns.

USAF F-16 vs AI dogfight, from DARPA with video


Army.ca Fixture
Fallen Comrade
Reaction score
This sure looks intriguing, further links at orginal:

An Air Force F-16 pilot will face off against an AI adversary in a simulated dogfight. Here's how you can watch it live

Mark your calendars, folks: The Air Force plans on pitting a seasoned F-16 fighter pilot against an artificial intelligence adversary for the first time in a simulated dogfight next week, according to the Defense Advanced Research Projects Agency, and you can watch the action live.

Between August 18th and 20th, DARPA plans on broadcasting its AlphaDogfight Trials that will see fighter pilots from across the U.S. armed forces face off against a variety of AI algorithms in "within-visual-range air combat maneuvering" — a dogfight.

The trials are designed "to energize and expand a base of AI developers for DARPA’s Air Combat Evolution (ACE) program," according to DARPA, a program that seeks "to automate air-to-air combat and build human trust in AI as a step toward improved human-machine teaming."

“We envision a future in which AI handles the split-second maneuvering during within-visual-range dogfights, keeping pilots safer and more effective as they orchestrate large numbers of unmanned systems into a web of overwhelming combat effects,” DARPA’s Strategic Technology Office program manager Col. Dan “Animal” Javorsek said in a 2019 press release.

You can register to watch the simulated combat, streamed live from the Johns Hopkins Applied Physics Laboratory (APL) in Laurel, Maryland, here until August 17th.


The Defense Advanced Research Projects Agency (DARPA) created the Air Combat Evolution (ACE) program to increase warfighter trust in autonomous combat technology by using human-machine collaborative dogfighting as its initial challenge scenario (DARPA photo)

DARPA originally held in-person trials in November 2019 and January 2020, with eight teams selected to participate in the third round in August: Aurora Flight Sciences, EpiSys Science, Georgia Tech Research Institute, Heron Systems, Lockheed Martin, Perspecta Labs, PhysicsAI, and SoarTech.

The third round of trials will see these eight teams fly their algorithms against three APL-developed adversary AI algorithms on the first day, August 18, before participating in a round-robin tournament to select the top-performing AI contenders on the second day, August 19.

On the final day, August 20, the top four teams will compete against an F-16 pilot in a single-elimination tournament to determine the winner of the "AlphaDogfight Trials Championship" — and, more importantly, determine whether an AI agent can successfully best a flesh-and-blood aviator in a dogfight.

“We are still excited to see how the AI algorithms perform against each other as well as a Weapons School-trained human and hope that fighter pilots from across the Air Force, Navy, and Marine Corps, as well as military leaders and members of the AI tech community, will register and watch online," Javorsek said in a recent statement.

"It’s been amazing to see how far the teams have advanced AI for autonomous dogfighting in less than a year.”

You can watch footage from DARPA's previous two AlphaDogfight trials below, or learn more about the competition here. I'll be making popcorn.


Not to take anything away from the work these guys are doing, but something similar was being done in the 1970's!

How Drones Beat Top Guns — 50 Years Ago
David Hambling


The Air Force has announced a plan to pit a fighter drone against a human pilot next year, reviving the longstanding argument about whether drones can ever match Top Guns and how good the Air Force’s Autonomous Systems are. But this same contest was held exactly 50 years earlier – and the humans lost.

The BQM-34 Teledyne Ryan Firebee, in service since 1960 is a target drone, a sleek, jet twenty-three feet long which could fly at over 700 mph on a pre-programmed flight path. It launched and recovered mid-air by a C-130 Hercules with a capture net, or parachute into the sea for recovery.

In the mid-’60s the military needed a more sophisticated target, one that would carry out evasive maneuvers like a real enemy. Ryan upgraded the Firebee with an Improved Maneuverability Kit (IMK) allowing a Firebee to make sharp 5G turns and banking at 75 degrees, making it a far more challenging target. However, the makers decided to go one step beyond and create something which was more than just a target. They developed Maneuverability Augmentation System for Tactical Air Combat Simulation or MASTACS.

Teledyne Ryan offered the MASTACS Firebee to the Navy. Flight tests took place at the Point Mugu Naval Air Station from January-April 1971. The story of what happened next is told in William Wagner’s 1982 book, ‘Lightning Bugs and other Reconnaissance Drones.’

Commander John C. Smith, head of the Top Gun naval fighter school , was interested in the new drone’s capabilities. Teledyne Ryan engineer Bruce Jackson promptly challenged him to a duel with MASTACS.

“I’ll bet you can’t shoot it down,” said Jackson.

Obviously, Smith took up the challenge. On May 10th as a graduation exercise, two F-4 Phantoms, one commanded by Smith and both loaded with live air-to-air missiles, took off on a mission to intercept MASTACS. The drone was remotely controlled by aerial combat instructor John Pitzen.

“Tally-ho off the left wing,” Smith announced on the radio, meaning he was about the engage — but the drone pulled a sharp turn before he could lock on a missile.

“It’s turning like a mother,” said Smith.

Smith lost the MASTACS drone, which continued to turn sharply until it had looped around and was on Smith’s tail. The drone’s ability to pull 6-G turns for prolonged periods which would cause a human to black out meant it could turn 180 degrees in 12 seconds at high speed. The unarmed drone lined up into firing position behind Smith.

The same sequence happened time and again as the two Phantoms vainly struggled to get the MASTACS drone in their sights. They simply could not keep in firing position long enough for their weapons to lock on properly. They fired two radar-guided AIM-7 Sparrow missiles and two heat-seeking AIM-9 Sidewinders without scoring a hit. If it had been armed, the MASTACS Firebee would have taken out both opponents easily.



  • download-1.jpg
    81 KB · Views: 44
But can it wear aviator glasses and swagger?

AI Slays Top F-16 Pilot In DARPA Dogfight Simulation
"It's a giant leap," said DARPA's Justin (call sign "Glock") Mock.


In a 5 to 0 sweep, an AI ‘pilot’ developed by Heron Systems beat one of the Air Force’s top F-16 fighter pilots in DARPA’s simulated aerial dogfight contest today.

“It’s a giant leap,” said DARPA’s Justin (call sign “Glock”) Mock, who served as a commentator on the trials.

AI still has a long way to go before the Air Force pilots would be ready to hand over the stick to an artificial intelligence during combat, DARPA officials said during today’s live broadcast of the AlphaDogfight trials. But the three-day trials show that AI systems can credibly maneuver an aircraft in a simple, one-on-one combat scenario and shoot its forward guns in a classic, WWII-style dogfight [emphasis added]. On the other hand, they said, it was an impressive showing by an AI agent after only a year of development. (As I reported earlier this week, the program began back in September last year with eight teams developing their respective AIs.)

Heron, a small, female- and minority-owned company with offices in Maryland and Virginia, builds artificial intelligence agents, and is also a player in DARPA’s Gamebreaker effort to explore tactics for disrupting enemy strategies using real-world games as platforms. The company beat eight other teams, including one led by defense giant Lockheed Martin — which came in second in the AlphaDogfight “semi-finals” that pitted the AI pilots against each other this morning.

Heron’s team did a live-stream Q&A on Youtube. “Even a week before Trial 1, we had agents that were not very good at flying at all. We really turned it around, and since then we’ve been really number one,” said Ben Bell, Heron’s co-lead for the project. The team intends to publish later this year some of the details about its reinforcement learning process for the AI, he said.

The trials were designed as a risk-reduction effort for DARPA’s Air Combat Evolution (ACE) program to flesh out how human and machine pilots share operational control of a fighter jet to maximize its chances of mission success. The overarching ACE concept is aimed at allowing the pilot to shift “from single platform operator to mission commander” in charge not just of flying their own aircraft but managing teams of drones slaved to their fighter jet. “ACE aims to deliver a capability that enables a pilot to attend to a broader, more global air command mission while their aircraft and teamed unmanned systems are engaged in individual tactics,” the ACE program website explains [emphasis added].

Heron Systems’ AI was extremely aggressive in the games, with its AI pilot consistently able to turn and score killing hits on the simulated F-16 piloted by an unnamed Air Force pilot, with the call sign “Banger,”  a graduate of the Air Force’s highly selective Weapons School at Nellis AFB. The AI exhibited “superhuman aiming ability” during the simulation, Mock said [emphasis added].

While the trials were not in anyway “definitive” of an AI pilot’s future capabilities or even its viability, Mock said, at the same time “what we saw was that in this limited area, in this specific scenario, we’ve got AI that works.”

DARPA intends to take the simulator used in the trials, and the simulations, to Nellis, where other Air Force pilots can take a stab at trying to beat AI pilots. Next steps will be to move on to testing AI pilots’ capabilities to perform other types of aerial combat missions.

Somewhere, the infamous Red Baron is no doubt laughing in amazement.

All fighter pilots look alike:

I noticed that too.

Literally isn't relevant, and nobody reading about AI fighter pilots dominating the skies really cares.  (In the context of the article)

Perhaps it's to show the industry & readers alike that companies like that have a fair shot at Pentagon contracts just as much as the big players?  :dunno:
Bonus points to whoever programmed the sim to read “flawless victory” on the screen after the shoot down. :D
Jarnhamar said:
Wonder why they felt this was important to point out.

Because they are both unusual in the industry.  The big players normally buy out / block via a variety of means smaller competitors.  And the military industrial complex has few women in positions of senior responsibility, in part because the military's systems for selection of senior leaders are weighted against them.  In the CAF, for example, the first real gate is officer DP3 training, where positions are not allocated purely on the basis of merit, but rather are weighted heavily towards occupational groups that have few women.  Blind selections of the best (without reference to occupation) would help counterbalance that, but would upset the apple cart and possibly result (long term) in fewer NWO, Infantry and Pilots in senior positions.
dapaterson said:
but would upset the apple cart and possibly result (long term) in fewer NWO, Infantry and Pilots in senior positions.

...and we can't have that, can we?  ::)

But yes, defence companies love to highlight the women/minorities/small company aspect to differentiate themselves from LockMart or Airbus, for example.
CBH99 said:
I noticed that too.

Literally isn't relevant, and nobody reading about AI fighter pilots dominating the skies really cares.  (In the context of the article)

Perhaps it's to show the industry & readers alike that companies like that have a fair shot at Pentagon contracts just as much as the big players?  :dunno:

Every Government contract RFP, RFQ and contract asks that.  You get points based on it.  In the US even more if Native American, Alaska based, Vet owned, Minority, Woman, HUB zone, Disadvantaged, Disabled etc.  It's in the first sections of the document. 
Apparently the PLAAF has been doing trials in simulators with pilots going up against machine-learning AI and are losing in dogfights as well.

Chinese Pilots Are Also Dueling With AI Opponents In Simulated Dogfights And Losing: Report

As an aside...since these AI machines are basing their learning on the tactics they are facing in their human opponents it seems to me that gaining (hacking?) access to this AI would give very valuable insights into the TTPs being taught to your opponent's pilots as well as the technical performance parameters for the enemy's aircraft that has been programmed into the system (as well as their assumptions about the performance parameters of our aircraft).