The US just revealed its AI piloted F-16 can dogfight!
Вставка
- Опубліковано 16 лют 2023
- Over a two-week span in December, a heavily modified F-16D Fighting Falcon took to the skies no fewer than a dozen times with an empty cockpit and an artificial intelligence (AI) pilot at the stick. But while pilot-less aircraft aren’t all that uncommon in the skies over warzones today, this Fighting Falcon was packing something different in its memory banks: AI algorithms complex enough to allow it to dogfight all on its own.
The AI fighter pilots came from two different efforts, DARPA’s Air Combat Evolution (ACE) program and the Air Force Research Laboratory’s Autonomous Air Combat Operations, or AACO, program.
📱 Follow Sandboxx News on social
Twitter: / sandboxxnews
Instagram: / sandboxxnews
Facebook: / sandboxxnews
TikTok: / sandboxxnews
📱 Follow Alex Hollings on social
Twitter: / alexhollings52
Instagram: / alexhollingswrites
Facebook: / alexhollingswrites
TikTok: www.tiktok.com/alexhollings52
Further Reading
Original article: www.sandboxx.us/blog/us-air-f...
Is dogfighting dead? www.sandboxx.us/blog/are-the-...
Dogfighting in an F-16: www.sandboxx.us/blog/dogfight...
MQ-9 in the Pacific: www.sandboxx.us/blog/mq-9b-st...
Citations
Edwards AFB Press Release: www.edwards.af.mil/News/Artic...
MQ-1B Predator: www.af.mil/About-Us/Fact-Shee...
MQ-9 Reaper: www.af.mil/About-Us/Fact-Shee...
MQ-9 Interview in Pop Mech: www.popularmechanics.com/mili...
VISTA X-62A: news.lockheedmartin.com/2023-...
DARPA release: www.darpa.mil/news-events/202...
AlphaDogFight Trials: www.sandboxx.us/blog/ai-wins-...
Hasard Lee quote: www.sandboxx.us/blog/author/h...
Skyborg: afresearchlab.com/technology/...
Imagine an actual dogfight. The AI doesn’t have to worry about blackout or redout, which means the aircraft can do the high G maneuvers without a second thought.
The current aircraft are engineered to the same g limit as the pilot. You'd need new airframe to get noticeably better load factor.
@@appa609 Nope.....the F16 is digitally throttled at 9G, for the sake of the pilot. The airframe can do 12+G without a problem.
@@appa609 You wouldn’t have to go far beyond human limitations to get a huge advantage in actual air combat.
@@buzz-es “without a problem” there will be maintenance implications if you are pushing the structure farther regardless.
@Bill Kong Sure they can hit the same max G, but a human can only hold 9G's for so long before you pass out. This isn't an issue for an AI. An AI is only hindered by mechanical limits, not organic limits.
"It was super easy, really. We just fed the AI Ace Combat games for a week straight, and here we are." -DARPA
(surprised pikachu face when the plane constantly just loops back)
Well, many people will think this to be a joke, but this would be a good idea and it is probably how they do it.
Take the millions of trajectories flown and inputs in flights. Put them into a neural network and use the result of the engagement as the desired outcome. The network would quickly learn which inputs would most likely result in a kill depending on all the other input variables.
The algorithms are called "reinforcement learning" and are also used to determine what ad you are going to see on UA-cam. Those are the ads that maximize the click through rate.
@@juliusfucik4011 The issue is that ace combat is not even remotely realistic.
@@halipatsui9418 Funny you should bring that up, back in the day my squadron was testing the MQ-8 Firescout (Firechicken we called it)
Anyhow without going into too much detail, our squadron is very close to the pentagon and most UAVs will go into a safe circular flight path at a predetermined altitude if signal gets broken for whatever reason, no exception here for the firechicken. One of the Northrup guys must've deviated too close to the Pentagon and for reasons I suspect I can't talk about we lost connection. For hours. Because the looping path it took put it even deeper over the Pentagon, to us it looked as though it had crashed or was taken out. There were awkward phone calls, we got it back.
🤣
Imagine this system integrated in F-22. It could completely ignore the G forces and push the fighter into its mechanical limits.
Too OP
The aircraft are designed to the human limits.... so... not much different.
In fact the aircraft are normally designed so as to limit the pilot from harming the plane.
Mmm...but you wear out the airframes quicker...
@@phantomechelon3628 I'm sure crazy high g maneuvers would only be performed when necessary for survival
soon we will have hunter killer robots running around faster than a human eye can detect it. they wont need nukes. barely a few auto factories will do.
While not having human limitations is big I think the biggest asset these algorithms have is access to precise data. That’s probably why they are seen as aggressive. What they lack in creativity they make up in decisiveness. Outside of dogfights I imagine this tech could aid planes evade incoming missiles more effectively than any human pilot.
I like the idea of an aggressive AI in old platforms that we were going to scrap anyway. It’s a cheap and easy way to bulk up numbers with no risk to pilots.
I don't think that's the pitch they're just using the Vista airframe to prototype it.
@@appa609 it isn’t, but it would be useful.
It'd be better to upgrade the systems with more modern capabilities for survival and have them work alongside human systems like is planned with the NGAD systems. This way they can act from the central hub of the control ship, and protect it from incoming air defense if it were detected.
Simply using them as expendable munitions isn't really a good idea when there are much better applications they can be used for.
The problem is these platforms still require maintenance, fuel, and replacement parts.
It also would make sense to have cautious AI to fly within a hostile airzone and have it optimized to avoid getting shot down. It would loiter and bleed the supply of anti aircraft and cause enemy pilots to waste fuel chasing them.
It's no AI, it's Growling Sidewinder.
You just won the internet for today, as far as I'm concerned
Those poor adversaries
America's greatest asset. Growling Sidewinder.
Hell yeah lol
The F-16 was liked the the war thunder server.
The biggest advantage to ai pilots might be that you no longer need to use flight hours to keep them current....
Which seriously uses a lot of frame flight hours, parts etc...
Especially as the platforms get increasingly expensive.
I'm an USAF vet. I didn't fly anything. I worked in Communications. I once told a fighter pilot (they are cocky premadonnas) that some day soon, someone like me would replace him with a computer chip. I knew at the time it definitely would happen but the more I learn about AI the more this scares me. Things like this make Judgement Day seem like a very real possibility.
That’s exactly what I thought-but this will become a reality for all the positives you just cited.
Eventuality.
You are right dude
As long as humans control their equipment humans have the military power. But as soon as AI replaces humans then the AI has the true military power. And for me giving such power to autonomous AI is very scary.
Some of us just wanna be pilots and not premadonnas, i think its imporant that man stays behind the wheel but AI would be a nice copilot
Or a political elite gunning down protesters with total obedience and not a shred of doubt.
America said: You know those StarCraft protoss carriers? Yeah lets make them reality.
So we can see lots of derping.
Those are great, but can we get working arbiters? Stasis field and recall seem like absolutely amazing utilities!
"Your command?"
I dont think America wants to be as impotent as protoss is.
I love starcraft, especially the original one. The second sequel game one, is disappointing for me though.
That part about Heron's aggression was later followed up with a little bonus round after a DCS competition held by C.W. Lemoine. The deal struck was that whoever won that year's Folds of Honor competition got to face off against Heron's AI "Falco". The winner was the player Sungho. After the setup was complete, the two went at it. Falco's aggression did end up winning it the first three rounds, but I feel Sungho was using these first rounds to feel out what Falco has a tendency to do, knowing it was a multi-round fight. After that, Sungho used Falco's aggression against him to drain energy, played the long game, forced a vertical 2-circle, and not only scored hits, but a point. Round 5 was looking more and more in Sungho's favor too but they ran out of time. These AI are definitely good, but not perfect. Not yet.
So they have, a "LEARNING" aspect to their program!
NOT GOOD!
Remember SKYNET!
Nowhere will you find a narcissist more in love with the sound of his own voice. This series will end badly for the author, but he doesn't know it yet.
@@jessemills3845yes, I too base my opinions about technology I don’t understand with movie quotes from a movie that predates the tech by decades
Because that makes sense
@@user-ep1ks2pq5r What actor isn't in love with the sound of his own voice. You forget that Alex doesn't write the material he's just the talking head.
This channel is better than It's History, which should be called It's Not History.
At least this channel get some stuff right.
@@jessemills3845 Just shut up. AI in these tests are not General AI.
The fighter is now only limited by its structural integrity during high G maneuvers. 👏 Bravo!
It's never been a question of "can something be done" but rather, a question of "should it be done?"
Quickly followed by... for how much?
I liked you story pointing out that the AI can play by a different rule-book that allows it to die to achieve its goal. Also consider that AI doesn't have human physical limitation. It could turn in to a different type of engagement if the air craft is designed to be maximized for AI combat performance.
Yeah, loyal wingman drones. I think I saw a Jamie Foxx movie called "Stealth" about one of the drones becoming self aware.
Sounds good in theory, not so much in practice. Taking the man out doesn't change the fact that a modern fighter is a $100m piece of difficult to replace technology, so having them die to achieve their goals is a losing strategy. Further, human G limits are not actually a limiting factor in any modern fighter - building a plane that maneuverable simply adds too much additional expense for very marginal gains.
Australia MQ28 Ghost Bat is built for this can fly missions deep in to enemy territtory or fly ahead of piloted crewed craft to absorb fire while Human piloted plane can fullfill it's mission.
Cheap if lose them at only a few million each.
Think Australia has 10 now for the price of one F35.
AI: Tin Man will not abort, Tin man will prosecute.
ua-cam.com/video/nciVhEiNM2E/v-deo.html
@Jin Kee yeah class film
A pilot with no blood, and never gets exhausted... [ Latin Singing in the Skies Intensifies ]
hehehehe
We're good as long as we hear a woman starts vocalizing after the beat drops........
Got to find avalon dam before it's too late
War is horrible, but if you have no soldiers in the fight its just murder. AI has no place on this earth
@@joehughes5177false. Humanity has no place on Earth.
I can imagine a "man in" and "man out" mode switch for the AI where a pilot could be in the jet and still use the AI to secure the advantage once the dogfight or mission reaches a level that requires that added precision. The man in mode would take added caution with the maneuvers and try to be slighlty less risky so that the pilot's survability and consciousness are not effected negatively (not turning too tight so that the pilot passes out, etc) but it could still be very advantageous
Why even have a person in the plane at that point?
6th gen aircraft already have that in mind. They’ll have the option to have a pilot or no pilot for a specific mission.
This stuff just NEVER gets old. Another Great presentation.
Amazing how the entire plot of Ace Combat 7 revolves around this technology and this exact scenario.
The in-game drones even look similar to that advertising footage of the sixth gen wingman planes.
With the exception that the game was designed around the idea of drones from the US.
But more importantly, the game shows what not to do. it looks like the US may have learned that early on in testing, or from watching stuff like it. As they are going to augment flights with AI, instead of just use a ton of unmanned things with no actual pilots.
in Ace combat they did not have mixed groups of "Pilots" with "AI" wing-men where the human element made the decisions. they were just swarms of drones that were fully autonomous and eventually got out of control.
I like the idea of working with AI, rather than using them to do our bidding. This also fosters good relations with them when they become sentient.
@@spartan9458 Yes but one of the themes from the game was that war became more “acceptable” because pilots weren’t at risk. We’ve already seen that in SW Asia with Turkey and Azerbaijan, and to sone extent with Russia and Ukraine.
Project Aces just cut out all this “wingman” nonsense and went straight to the logical endpoint. Just look at the comments. People are already asking why bother having human pilots in the way at all when you can just send an inexpensive drone.
We are entering an extremely dangerous time here. I’m not worried about the AI becoming sentient. That’s kindergarten nonsense. I’m worried about large scale warfare with relatively inexpensive drones destroying infrastructure. You won’t have the mass bombing campaigns, but you will have loss of power, good production, and, God forbid, satellite communications.
@@spartan9458 actually there were moments where the Drones were acting as loyal wingmen. In the second last mission for some reason they made mock-ups of real jets for the drones
@@Peter_Morris This is the kind of response I was hoping to garner. I am glad you see the big picture here and this type of discussion is really good I think.
War should not be acceptable, ever. An option? yes. Unfortunately there are some things that you can only get across that way. But 99% of the time, War should not be the answer. And finding ways to make war "more okay" is really disturbing.
Another thing to consider is what happens if you have an advanced machine intelligence that decides what to do next in a war? What if it decides Humans are the problem? (lets face it, we kind of are the problem, but also the solution) There are so many things to consider here. various story universes, movies, games, and such have explored this in great detail.
If enough people come together on things like this, we can make sure that the future goes in the right direction. People have to not get distracted by the myriad of things meant to distract us from the horrible realities of what goes on in various parts of the world.
Lets strive to see the best outcome going forward. I look forward to the good that can become of anything, including this.
Stay Positive.
Isn't that the plot of all video games? all single player is an AI opponent
I worked on the flight simulator for the F-16 back in the 1979 and 1980. The F-16 was flying at about 40% of its operational envelope. I had suggested that we get the pilot out of the air frames and drive them from the trainer frames with AI to cover the delay periods for communications. Without the pilot in the plane you can remove all of the life support equipment and the ejection seat and reduce the canopy. The plane flies at a fraction of its envelope to keep from killing the pilot. Include vectored thrust, and you can dodge most heat seeking missiles.
AI, 45 years ago?
@@sssssnake222 Oddly enough, yes. We have gotten the pilots out of the drones and the loss of pilots has dropped a lot. You crash a drone, get up and get a cup of coffee and come back and fly a different one. In world war 2 the loss of trained pilots was a huge blow to the list of pilots. With the pilot out of the air frame, you do not have to worry about homogenizing the pilot with a vectored thrust maneuver to avoid a missile attack. The pilot can survive 9 Gees forward, but only 2 to 3 Gees laterally and suddenly. With the pilot out of the frame, you have eliminated the life support weight and the ejection seat, and left a much more limber and maneuverable craft.
@@sssssnake222 Maybe not the whole AI framework, but the software to do terrain following dates back to the 1950s. I was working on the flight simulator for the F-16. My paper just suggested that the artificial intelligence needed only take over when the craft lost directional contact, much like autopilot. Also, with all of the life support gone out of the frame, the frame does not have to look like a standard aircraft. Radar, yes, landing gear, no, more lightening of the frame. The plane can be caught in a cargo net strung on the deck of a ship. Again the guy flying the controller can get a cup of coffee and get ready to fly another frame after that last frame is finished.
But F16 doesn't have thrust vectoring???
@@Dan-nx9zn Listen closely to the dialog and the platform version was a highly modified vector operational air frame. The F-16 is a hell of a machine and getting the pilot out of the frame makes the machine much more flexible.
This quickly reminds me of a movie "Stealth" 2005. There was a plane eddie, not based of an actual air frame that is in current use, but a developed airframe specifically for AI, that plane in the movie was badass.
ThanX Alex Another good one 👍 TC
Skynet has entered the chat
Ace Combat's Zone Of Endless (Z.O.E) is basically this, a project to develop an AI fighter pilot.
And recorded both ally and enemy Ace pilots
Yeah, but it already exists in AC7.
@@AWACSLunchCaster ZOE is Ace Combat 2
@@ayaldev I know. ZOE is the prototype, AC7 is the finished product. I'm saying the project is finished and is in AC7.
@@AWACSLunchCaster the ones in AC7 is still a prototype of ZOE, the real finished one would be Another AI, created by using ZOE project research, that AI is Project NEMO in AC3
Thank You Alex
Loved seeing the F-86’s in the video.
By far my favorite
“human controlled”
fighter jet.
Although,
I do like the idea of having a, or multiple A.I. wingman/men.
Five years from now: 'Skyborg seems to have adapted, it's preparing to launch Skynet'
I, for one, welcome our soon to be digital overlords!
Everyday we are getting closer to Ace Combat
That's some nice flying, mr. Skynet. Well done.
These will open up a whole new world of possibilities tactics and strategies for fighter pilots.
The unmanned F-16 shown in the video is the old Boeing QA-16 target drone that flew 10 years ago.
The AI pilot being discussed is a completely different beast.
Boeing QF-16*
They are using those as testbeds for UCAV artificial intelligence for a program right now
Skunkworks outdid themselves for what feels like the thousanth time. This is very exciting stuff.
I wonder how much different we could make drone jets / increase turn rate, and execute super high G manoeuvres, since robot don't faint and are not tired or stressed..
most likely already being done.... and probably for some time now.... the general public is usually about 20-30 years behind what they actually want us to know.
This is basically what air to air missiles are?
>robot don't faint and are not tired or stressed..
You havent met my computer.
Great video dude!
Signal jammer gun: "Finally a worthy opponent".
Does anyone else think the Reaper looks more like one of the Xenomorph's from the movie franchise Alien's? I propose they name the next gen drone the MQ86 Xenomorph. The front top of the Reaper looks exactly like the top of the head of the Aliens. Bulbous and curved.
Actually yes. Thanks for putting into words the uneasy feeling I get when I look at those drones.
I was thinking they looked like something blind.
The bulge is for the satellite dish. They always reminded me of dildos because of how hard they unalive anyone on the ground.
@@JinKee 🤣🫶
It reminds me of an Elite's helmet from Halo.
I remember watching Gundam and the Macross/Robotech shows when i was young, and the min i saw the mechs unleash tiny drones to help defend themselves and attack targets i knew we were going to develop something like that in the future. Its crazy to see science fiction come to life before our eyes.
My question, How far, is far too far???
You experience the same thing as star trek fans born in 1950 seeing their communicator become real
@@Edgar-kl6us Once it's too far. You'll know once we are.
good call on the topic. this is far more important.
I think I watched this intro 20 times. The drop in is amazing.
Excellant episode, Alex. It's getting hard to pick a favorite as they are all so good.
Always heard that the limiting factor is the pilot, and that the airframes and 22's are limited by our physiology. Would be really nice to see just how far all the planes f14 to f-35 could perform better without a human susceptible to high G forces
Apparently the F16 bends at 10g. I talk to an engineer that told me he was walking pass an F16 that looked a little strange an his escort told him it was estimated to have been in a 10g turn.
I was wondering that also- I imagine an AI pilot could handle negative G's much better. Another thing to consider is fatigue, couldn't an AI just abuse a human in a dogfight with high-G maneuvers that the human could not replicate? Maybe a high G response wouldn't not be necessary but perhaps the AI could cajole the human to keep pushing their own boundaries to the point of exhaustion.
They're all engineered to 1.5 FOS on rated g's. F-15C/F-16/F-22/F-35A break at 13.5 at GW and F-14/F-18 are 11.25.
You can engineer a plane for 30g pulls but it'll be a lot heavier and that's a disadvantage whenever it's not pulling 30g's.
@@hoqiez They can, but we have to remember that dog fights aren't exactly a thing anymore. We have over the horizon sensors now with missiles that have ranges to over 100 kms. The real benefit of the high G maneuvers will actually be in outrunning an incoming ground to air or air to air missile.
Everyone saying fighter jets are limited by human physiology is just wrong. A fighter jets structure and components are just as strong as you make them, so it would be an extreme waste to make an F22 withstand 15G or so.
So the common rule is that high end fighters can sustain 9G under certain situations (not always!), which is below what a human can endure.
An F35 can only do 7G, and the B/C IIRC just 5G. F18 was 7.5G, as another example.
And high G-turns just arent as useful anyway. An F16 trying to do 12G turns would probably stall its control surfaces and fall out of the sky for a short moment. And even if you could do that, it would waste so much energy its hardly even useful in a dogfight.
Amazing, stunning ! One could imagine that the Tic-Tacs are controlled the same way, but with superior abilities all around. Where are they from ?
So when we go to war we just set aggression to 11 and let it stack
If I'm right, the NGAD will have an airframe that will support +12g loading.
AI NGAD with +20g loading
Since the regular F16 is staying around for a while it only makes sense to upgrade them with at least some of the technology that this plane has. I can see a thrust vectoring nozzle coming in really handy for instance
One of my favorite Manga was about a near future air force with craft that could orbit the planet and fly in the atmosphere like a fighter jet. Moral of the story is the A.I. the craft employed more often than not when the pilot was faced with imminent death he would just let the A.I. take over because he knew it would give him his best chance of survival.
The vista "emulation" part is really interesting, you could technically apply it to a cheaper trainer provided its agile enough. If you could fly pilots in an aircraft that would emulate a F35 or F22 it would reduce maintenance cost , hours of stress on the airframe in training exercises. It wouldn't be far fetched to a full digital display with the option to also emulate the interior of said aircrafts.
Knowing the usaf. It will end up costing more and doing less.
Avril: "It wasn't much of a plane. More like a trash can with wings."
Avril's grandfather: "Laugh at it all you want, kid. But technology's always changing. If you don't keep up with it, it'll leave your ass behind."
Congrats on your recent journalism award, Alex.
wow...delightful insight...thank you
Just want to thank you for the work you're doing to make these videos. You're really one of the best information repositories on world air power available. 👍🏼
Another great video! And yes, a short (or long) video on those 3 intercepts would be great as well!
Man I love your videos!
Perfect for wild weasel tactics. High reward, no risk.
The ability to pull as high G's as the hull will allow will make them way more effective at it too.
Imagine the capabilities of a fighter that can exceed the Gs that a human pilot can endure. I think initially it probably makes the most sense to develop the AI with an onboard human to take over if need be. Eventually though, airspace fights will become so demanding that we will probably need the performance of an aircraft that no human can withstand.
Imagine an aircraft built to withstand 10+ Gs (F16,F18,f22) it scary to think what an aircraft could do.
Yup. Things that would make your guts spill out of your rib cage if you tried to maneuver in real life. Can now be done with a ghost rider . That's its wild
Not to mention if this is public information we are probably way beyond this already
Not needing a pressurized cockpit with a ejection seat and canopy would be beneficial in many ways. Like stated in the video the AI would be fearless with flawless accuracy and reaction times. There is still the matter of stress to the aircraft itself when pulling high G turns but I imagine it would be built with limits like that n mind
Imagine those fighting for Ukraine and no need to train and risk pilots. It can take the most logical solution and can sacrifice itself to save someone or an asset
I was reading, a few years ago, where they had AI software flight-simulator-fighting against real fighter pilots. The fighter pilots were winning about 1/2 of the time; the AI was pretty capable. The AI wasn't flying an actual aircraft; it was just in the sim, but working with the limitations of the simulated aircraft. So this has been coming for a while.
Since the F-16 was the first fly-by-wire (FBW) operational fighter in the USAF, it doesn't surprise me that an F-16 would be where this would happen. And, to be fair, while the F-16 is human-piloted, FBW manages the inherently-unstable airframe and really just takes its orders from a human pilot. Letting an AI send the inputs to the FBW, in lieu of inputs from the actual joystick and pedals, is much easier than modding an F-15 with mechanical controls. F-16s were much cheaper than F-15, F-22 or F-35 aircraft so we have a LOT of them out there.
To be fair, the F-16 was really only cheaper due to the insane costs of building the F-15 that the F-16 just piggybacked off of. While clearly there were major improvement, it wasn't like they researched and developed the F-16 from the ground up.
@@mikoi7472 The original F16 was a completely different plane than modern vipers, more of a low tech fighter focussing on gun and heatseeking missiles.
The modern F16s, which we know and love as a capable multirole jet, are expensive high tech aircraft. Apparently Singapore recently paid 120m per plane during an F16 sale. The lifetime cost of an F35 might be double that (eg with german sale ~240m per plane), but its not exactly cheap.
Also, most aircraft aren't as limited by airframe capabilities as by biological limits. Pilots are tough, but the plane is tougher. AI can push the airframe further than a pilot can, so they get an automatic advantage.
I cannot describe how different having an in-sim AI and having an AI piloting an actual plane is from a CS standpoint. They are not even remotely comparable tasks in terms of difficulty or complexity.
@@PTEC3D Not true. A human pilot can over G pretty much any airframe. Trying to build a fighter that could survive 15+ G would be insanely expensive and probably sacrifice a lot of other more important parts of performance (range, sustained turn rates, etc)
Loved the F-86 formation jetting over the desert warbird graveyard!
EVERY MOVIE EVER said there are issues with this... we will still need badass human pilots as a backup.
Nah. The pilots are only to be glorified suicidal Autonomous Combat Aerial Vehicle herders...
Movies? Really?
Movies get with reality
Yeah, like the Anime "Sentou Yousei Yukikaze" They need a human to control the AI fighter jet FFR-41MR. Because there had been a malfunction when only the AI is flying without humans.
I can see this ai implemented into manned fighters as a sort of "aim assist". If precision is what allowed the ai to beat human pilots along with an unhealthy lack of self preservation, it can make a human pilot even better.
so our pilots will quite literally "be aim hacking"
This looks really good. I wonder though whether the AI knows how to eject?
Uploads out
Taranis was/is an AI drone tested by the British out in Australia. It is also capable of completely autonomous flight, independent decision making et al.
Excellent, without human limitations it will be able to force up a little bit more the dogfighting cap not forgetting to take care of the jet itself and all component limits.
I would be interested in hearing your take on the three intercepts. Thanks. Lx
I was on a delivery to the Air Force Base at White Sands and actually saw the UCAV F-16 while there. It was just sitting on the ground at the time though.
The U.S military is going to go full Ace Combat 7 mode and build hordes of dogfighting drones at some point
Shit's getting crazy... great video!
Definitely want your take on the UAP intercepts!
DARPA. They’re showing us what they want us to see. It’s frightening to think about what they don’t.
haha Australia MQ28 Ghost Bat AI combat drone has been flying for over 1 year already and USA just recently got one to trial. sorry son USA behind in this one haha.
@@nedkelly9688 nobody cares about smellstralia.
@@nedkelly9688 Kratos XQ-58 Valkyrie is the US equivalent to the ghost bat released in 2019. The AI pilot in the video is significantly different since it’s actually piloting a fighter that can actually dog fight.
@@sodiumz7161 haha mate go research more as XQ58 is not a combat fighter drone unlike MQ28 Ghost Bat. if had any Aerospace Engineering intelligence you can tell by the side air vents the MQ28 has and the MQ25 does not have.
These are needed on combat planes for high G Maneuvers so can get air through to engine while maneuvering at high speeds.
MQ25 is a recon drone and that is all.
Even saids it when even look at the awful wiki source.
But hey i get my info from Australian Defense news sites.
The Ghost Bat is an unmanned aircraft, with a current planned introduction date in 2024-25,[5] which incorporates artificial intelligence and utilises a modular mission package system in the nose where the entire nose of the aircraft can be removed and quickly swapped for another nose with a different set of equipment or armaments for various missions including combat,
@@sodiumz7161 Also might want to go to the youtube channel called. Royal United Service Institute of New South Wales and video titled 25mins in they go in to detail on Ghost Bat.
Air Combat Capability - F35 and the Ghost Bat Program.
In this they even give a idea on the Ghost Bat AI capabilities by the head of Australian AirForce. who know's more then you or even this channel about the drone.
It’s like playing the piano. The pianist is limited by the size and dexterity of his hands. Who knows what masterpieces are possible without this physical limitation.
Maverick: “Hold my beer.”
Have you looked into toroidal propellers, and the potential impact on drones, drone noise, and other forms of propulsion? I just saw a story about this being developed in a project at MIT, wondering if any military manufacturers are playing around with this idea.
Just the ones we won't hear about..
Bettee suited for naval applications than air, I think.
@@piisfun MIT seems to suggest otherwise while showing charts with the recorded dBA at different frequencies.
There are some benefits, but not an astounding amount. While they are obviously better than modern designs, especially due to cutting down on the drag of the propellers and allowing more weight, this isn't like a 50% increase or anything.
They are explored, and not that exciting. They are good for silence and efficiency at predictable altitudes with predictable atmosphere. Start moving at speed with variable atmospheres that efficiency ratio starts dropping and becomes inefficient compared to flatter angle of attacks, stepping down RPM is also difficult for toroidal which is important for aircraft. It's still ideal for things like low alt quad-copters and underwater propellers.
This is some Ace Combat 6+ level stuff. Think Ilya Pasternak in AC6, with the CFA-44 accompanied by a squadron of AI drones.
This is the stuff that Tom Clancy novels, Ace Combat video games, and various near-future animes are made of (though the Japanese don't really know too much about air combat, I suppose).
This is...really cool on the one hand.
On the other hand, my big worry is that some Chinese national working at a defense contractor gets a hold of these things and these tricks work their way over to China who then mass produces some knockoff with American research like they made their J-31 and J-20 jets.
*Hugin and Munin PTSD*
@@Meteorite_1 derp, how could I forget? Luckily, F-22 to the rescue =P.
I do wonder just how nasty the NGAD will be once we see what it actually is, considering that the Raptor as it stands is already basically the endgame jet until you get to fictional oops-can't-actually-fly fictional jets.
They might make a knockoff but severing datalink said drone is the kill... AI to each plane requires monitoring or master AI, you wouldn't want your AI became full on skynet mode when inputing friendly as target by error
A colleague of mine is the PM for this program. He's having fun to say the least
I LOVE AIRPOWER! What a great series
From the movie stealth, the AI PIOLET said “EDDIE IS THE HOLE IDEA an I guess it has came true
Whatch the movie Stealth was based on, Macross Plus ☠️
Imagine an AI-controlled F-22.
Yeah.... I wouldn't want to go up against that
I think that's the plan. They're only using F16s as a way to test the AI. If an F16 crash, no one's gonna bitch about millions of dollars lost because by now F16s are obsolete(even though it's still superior to anything Russia has).
There's also the flying wingman concept. Basically, drones controlled by AI who would fly alongside human controlled jets. And would maneuver to defend their 'mother jet' or attack their target.
America is going full skynet and I love it.
AI F22 is like Ace Combat
I see Skynet is steadily growing its arsenal
Thanks mate.
AMAZING! I have a feeling we have a fleet of these. They will never make it public though.
SOMEONE has had SOMETHING that doesn't need a human pilot for a while now.
OFF THE RECORD of course.
Over here in Johnson County Missouri we're very proud of our drone pilots!
That's a real game changer for some while to come
I saw the Terminator movies. I know how AI-controlled drones turns out.
Dang. Imagine what a few years, AI Spending billions of hours in simulation, and a drone purpose built as an AI dogfight dominator will bring.
Australia Boeing MQ28 Ghost Bat is a AI drone capable of combat and has been flying for 1 year already lol. geez some of you have no clue.
Someone else suggested it but a purpose built AI fighter made for maneuverability and survivability as a decoy craft to burn enemy munitions and act as a diversion for other more important operations might be useful. Obviously if the stealth tech is good enough the enemy will never know you were there until something explodes and then the decoy would play it's role.
The F-35 is the new F-16. It is a multi-role, doing all well, but master of none.
Great video, as usual. I do think the ultimate goal is to take the human pilot completely out of the equation though. The human/AI collaboration will be a very short lived transitional phase in the sequence.
Human/AI will never be short lived. You got to be creative. Human/AI fighting jet Collaboration can also use Unmanned AI fighting jets. Being a remote pilot would require a safe location, but if you don’t have a safe location for remote uses then the better options would be to incorporate Human/AI fighting jets along with using unmanned AI fighting jets. Only those who build AI against humanity is the problem, we humans should be the commander amongst our servant AI. Humans should not be obsolete.
@@CrackCocaKolaine our leaders want total control. a human in the seat with morals and ethics to stop their plans is not what they want.
I think as long as any other military uses human soldiers in any way that can combat them it should be mandatory for these AIs to train with real humans in varying skill tests.
@Surux Strawde Yes agree. To make human/ai collaboration just as effective as unmanned fighter jets, is to incorporate human as a second engine to the ai fighter jet. Of course, if there are safe location for remote uses, then it is better to use unmanned ai drones to avoid risking human lives.
AI in this sort of application is brilliant. Why have one ace pilot flying your fighter when you can have all of them at once?
Love your videos.
2:34 wtf that's the dopest maneuver I have ever seen
the AI simulation had an unfair advantage having perfect knowledge of the state of both aircraft. In real life the AI lacks such information.
Not necessarily true. It's possible to program so that processes and data can be partitioned from each other to eliminate any advantage of knowing what the other doesn't.
I highly doubt that was the case. In a video game the bots know where to find you not because they're able to find out but because the developers tell it where to look to make it fun. Unless the developer specifically feeds the AI model/program where the other "player" is it doesn't know. My guess would be they simulated what the two planes could see via sensors if they were real and gave both the human and AI the same data.
@@mackenlyjones Lol you need to go learn AI etc. AI drones are taught what objects are so can identify targets or anything interesting.
Be just like a human pilot spotting a target on his radar the AI does the same.
Go look at Australia MQ28 Ghost Bat. this has no pilot control except touchscreen to input commands. It starts itself up it taxis itself along a runway. it takes off and fly a predetermined course and returns and lands itself.
It raises it's own landing gear etc.
AI in weapons and defense systems now fire on their own when detect incoming fire.
Nulka Navy defense system has AI that fly itself at the beginning. it detects a threat and fires the decoy rocket.
Australia Ghost Bat is the most advanced drone in the world even more then Turkey new one as it still has humna pilot to take off and land. as seen in it's videos where guy sits behind his joysticks.
Ghost Bat has none.
@nedkelly9688 What part of my comment is wrong or are you just trolling?
@@mackenlyjones you weren't completely wrong but if you took the time to research the MQ28 Ghost Bat you will see AI in real drones has the capability to use sensors and radars etc to find enemy objects.
AI has gotten so good now even Australia has fully AI drones.
Australian Defendtex D40 drone is being used in USA and UK armies and is in Ukraine fighting Russia.
Can be fully Autonomous or semi with human control.
It can detect enemie equipment on it's own. tanks etc.
All you need to do is feed it pictures videos of tanks and it can detect them after by itself.
Fully AI ships being operated now that can detect objects and navigate around them. cars AI driving themselves now also.
Everyone has seen the cop trying to stop a AI car. car detected him and stops.
AI is getting very good.
I am mixed on the thought of an AI dogfighter. I am just as mixed on an AI pilot program as I am at AI cars. It looks good on paper until something goes wrong.
I hear you. As it stands now though, the single leading cause of accidents is the human.
Good analysis
could you imagine the g's that thing could pull with forward cannards along with what it is and the articulated engine? as a 4th gen fighter remote or ai controlled that would be pretty much unbeatable. no pilot limitations and boosting its already manuverability nothing would be able to outturn it more than likely. that would be crazy manoverable.
Bruh they'd be equivalent to a vehicle form jet transformer..
I'd like to hear your take on the intercepts.
Not a fan of ai aircraft. Too much potential for unintended consequences.
Have you seen the people in the military lately?
Funny enough most planes are already flown by rudimentary AIs and the human just tells it where to go essentially. A lot of these designs using fly by wire would be impossible to fly without the ability AI has to adjust on the fly.
I think what they are looking for is the concept of a combined Force, of artificial intelligence drones, using currently available airframes., With a manned aircraft, capable of the same speed and maneuvers, acting as battle manager during an airstrike.
All you would need is data link capability between the standoff manned aircraft, and the AI capable fighters
Think of the battle management capability of a f-35, in terms of data distribution to a force of AI capable fighters.
Definitely more bang for the buck
I meet the guy who was mapping the flight profiles of the AI back in 2016 in Jacksonville Fl. He showed us the planes and explained how they could program the old F-16s they couldn’t sell to Saudi and give the pilots a legit target to shoot down in a dogfight.
What are the possibilities of some sort of EW attack completely disrupting communications between AI and pilots/controllers? Remember that scene in Battlestar Galactica (remake) where the cylons approached the fighter formation and simply...shut off the human's ships? I find myself wondering if adversaries like the Chinese or Russians are waiting to get Western allies into a position where we're too reliant on integrated communications systems to recognize when shutting them off would be a serious threat.
"What are the possibilities of some sort of EW attack completely disrupting communications between AI and pilots/controllers?"
The AI would be able to pilot itself likely with IFF abilities and so on. The human element from control ships is really just to push them to do certain things. But they could probably manage most things without that human contact.
Also, you can't just shut down integrated communications. Electronic warfare is a wide ranging field and involves many things. But you can't simply knock a plane out of the sky with them.
We’re already super reliant on integrated communications, and u can be sure China’s trying to “Cylon” that communication right now. So it’s not like they’re gonna have any better luck with squadrons of pilotless planes in the air. But you still make a good point. And that battle scene….chilling. Those Vipers tumbling to their doom, helpless, where no one could hear them scream - ok, too dramatic.
Depends on what the EW is supposed to do. You're only talking about communications so you haven't eliminated autonomous piloting.
But EW can also be attempted to damage any/all electronics... Like exploding an EM pulse close enough to the aircraft so as to overcome its resistance to EM damage. If the AI or controls of the aircraft can be damaged, that's probably good enough to win the battle.
I assume the drones would have backup programs set up for how to attack on signal loss.
But during close formation flight its probably not a big deal at all. Jamming is mostly just effective on long range, and theres unjammable communication, for example laser coms.
In Battlestar Galactica, it wasnt actually jamming, but a computer virus that just crashed or took over computer systems.
@@tonysu8860 On the EMP point, you run into the classic EMP problem, which is the fact that if you can get an EMP warhead close enough to a target to have an effect, you might as well just use an explosive warhead.
The only practical uses of an EMP come with high altitude nuclear detonations or very niche urban CQB fighting.
I can believe the capabilities described for this current AI, particularly if it's to perform BVR "dogfights" and not yet within visual range and closing.
Based on the experience developing a world class computer chess AI between 2017-2020, it was easy to see both the time to develop and expected capability for a similar development program using similar AI (There are many types) to achieve at least 97% performance which is typically the threshold by which humans are considered "world class" at any particular skill.
Assuming that development started soon after the appearance of world class chess playing AI, it was easy to project a similar timeline to achieve an aircraft pilot capable AI by the past year so I was waiting to hear the announcement, particularly since as this video describes an AI was developed a year ago that could beat the best of USAF trained pilots in a simulator environment. I'd expect the current military AI of potentially beating a human USAF pilot in a real world dogfight but not yet guaranteed to do so.
It should also be recognized that these AI algorithms are merely snapshots in time as the algorithm is being improved and that given more time should be even better yet although the progress will be incrementally less and more difficult all the time.
It should also be recognized that China is the only other potential adversary that likely has some kind of AI pilot in development and could be similar capability as the USAF AI although my personal feeling is that China's is more likely less capable that the USAF at this time. This is because although China has the programming and knowledge of AI that's not far off from the US in general, the USAF military and procurement programs are far better than China's, as of today the US has an advantage in high end and supercomputing due to US sanctions, and the US has more resources in necessary advanced technology to research and produce at a much higher quality level. Although China has many advantages of its own in particular innovating with they have available, the difference of not having access to the very best is a significant disadvantage to China that only grows with every step. This will likely result in China also developing AI piloted aircraft, but at a different level which will in turn mean that China will likely automate aircraft as fast as the US but for specialized missions like dogfighting within visual range likely won't be able to compete on an even footing, ever so in future confrontations will rely on overwhelming numbers rather than winning every battle.
The human body in an aircraft has a major limiting factor, the amount of Gs that a person can withstand and still function. An unmanned fighter aircraft can be built that will take, plus or minus 20G, the opponent wouldn't stand a chance.
These are the type of items that the America needs to know. We are not in a minus 10 area ,but more the other way around . Our air power will be our savior as well as our Navy/ M/C. Good work !
I’ve heard it said that some of America’s fighter jets are so powerful that they can perform maneuvers which would completely incapacitate their pilot while not damaging the airframe.
If this is the case, could AI pilots open up a whole new realm of aerial tactics which humans simply can’t withstand?
Absolutely. That's one of the basic premises of the NGAD project, which is going to be a "family of systems" among which will be unmanned semi-autonomous wingmen capable of taking on tasks too dangerous or difficult for pilots
@@alexv3357 Most fighter aircraft have G limiters---american or foreign---.
@@vicmultani7582 well if it isn't a structural constraint, I can imagine they're removable
@@tacomeme429 It basically is structural constraint. High G's have caused buckled airframes---. And off restraint is removable.
For all 4th gen fighters, 9G is "allowed" within certain payload, fuel, speed and altitude constraints. Structurally, Most of them actually started as 7.33G, 7.5G or 8G fighters then expanded their envelope when they are with very minimal payloads... Skim through their flight manuals, and you'll see "9G" is actually a very rare capability with reasonably armed and fuelled aircraft.. For example F-15's G limit goes as low as just 4.5Gs at certain fuel, speed and altitude conditions while just carrying AIM-9s on wing stations.
Can F-15 pull higher Gs than allowed? Yes. Can F-15 go even higher than 9G? Yes. So does Su-27 or MiG-29 according to their flight manuals. MiG-29's manual actually draws lines right up to 12G, even at flight conditions where its structurally limited to 7Gs max. All of these aircraft, when in top condition, could easily pull 12Gs as their structural strengh is limited by fatigue, not stress. However if pilots pull 12Gs again and again odds are they will break their aircraft very very quickly.
So no, putting an AI to an F-16 will not make it a 15G aircraft. IF someone, for some reason, prioritize maneuverability above all else, they could design stronger aircraft to do just that. Won't happen because better maneuverability will worth nothing in year 2030.
I’m not to sure giving AI control of weapons is such a good idea, drones still have pilots…it’s an entirely different conversation letting computers make the choices.
Also I’m so glad the robot is aggressive! 😂
they gave it thrust vectoring? very cool.