A lot of it boils down to capital cost vs. operational costs vs. capabilities (overall and specialized) vs. other considerations like pilot safety. See also just about any 4X game.
A missile isn't generally reusable - they're designed to operate at a region of performance within their engines that would not even allow for sustained flight.
If it only has to "live" for a few minutes, you can get a LOT more performance out of it, which is exactly what they're designed around.
But it's a trade-off. You could invest in a (practically) infinite supply of global capable missiles to be launched from the literal center of America, and to get 24 hour coverage you just launch one every 30 minutes. But that's extremely expensive and stupid.
But a missile which is carried to be closer to the battle can be smaller, and cheaper. Then you need some kind of vehicle to get it there. This kind of thing is that vehicle.
Lets just hope the entrenched pilots don't find some way to handicap the drone's ability to keep it "fair". Why the US is spending a trillion dollars on the F-35 program is an exercise in pork politics.
100% think this is the way this is going to go. For a year or two.
Finally the fly boys will realize they're totally outmatched and that the F35 will be our last fighter. In my wildest dreams it gets cancelled and a bunch of smaller defense companies spring up to bring prices way down and competition in the space increases. Then overall defense spending goes down and we get more money for services - while being still being armed to the absolute teeth.
The USAF will do everything in its power to keep human pilots in the loop. It would take a credible opponent (Russia/China) switching to drones and/or a third world oil country putting up a real fight with drones to change that.
Stealth is a 2005 movie that explores this concept. It didn't do too well in the box office but I definitely enjoyed it although the AI is a little forced.
It looks like the drone in question here is the following (quoted from Aerospace Testing International), also mentioned in the linked source:
'The “fighter-sized” 5th Generation Aerial Target (5GAT) is 12.2m (40ft) long, a 7.3m wingspan and a maximum gross weight of 4,350kg (9,600lb). It is designed to be launched and landed using a conventional runway. The drone features two afterburning jet engines and a 95% carbon fiber airframe.' It seems to be designed specifically to stress test our own flights as target practice and doesn't seem like it's actually going to be going into combat anytime soon.
If you enjoyed that, be sure to check out Macross Plus. The rivalry is between a traditional pilot and pilot who controls using a neural interface. A little less hand wavey than what sci-fi thought AI was 15 years ago.
I love how Macross Plus was inspired by the real-life rivalry between Northrop's YF-23 and Lockheed's YF-22, competing in the Advanced Tactical Fighter competition.
I have a hard time getting into animated stuff but I'll definitely try to download a decent quality version of it and watch it. Sounds like its right up my alley! Since this was created in 1994 and Stealth was created in 2005, I'm wondering how you arrive at your third sentence! Haha :)
I was referring to Stealth, which attempts to depict AI 15 years ago.
Macross Plus being much older doesn't attempt to get into AI for the opponent, and instead uses something a little more down to earth at the time which was neural interfaces.
"[AI] would be able to make key decisions faster and more accurately"
That's a statement nobody here will have problems to accept right? I expect nobody can think of any example that makes "more accurately" a problematic claim.
I don't think that's the case at all with flight vs vehicles. Autonomous driving is complicated because you have to take into account a wide range of scenarios and wide range of human behaviors, and a complex environment to understand. And then make very fuzzy choices on how to react to those scenarios.
Combat is a simpler zero sum game, pick and target and destroy.
Based on how AI plays top games right now(Starcraft and Dota), very fast quick reflexes and pin point accuracy of their attacks I think the AI drones are going to crush humans.
IMO the stakes are different. Often quoted is self driving cars must be 10x safer then humans before society accepts them. For drones we'll take 10% worse performance, we are OK to some extant of misfires particularly in a remote village in a 3rd world country. If drones cost 1/10 of a plane it will be a obvious answer for any general.
It seems like an entirely valid statement. AI can certainly make faster decisions than a human and it can utilize accurate 6DoF calculations and multi-sensor fusion that humans can only roughly approximate in an intuitive way during battle. I'm sure there will still be areas where humans will be better, such as analyzing complex visual information or predicting high-level reasoning of an enemy pilot, but I see human-versus-drone encounters as a clear loser for human pilots in the relatively near term.
Turns out humans are fallible, even expert one. If we assume a very, very tiny chance that a trained pilot will misidentify a target, multiply that by the millions of hours of pilot flight time in the US military, and out pop some rare extreme outliers on a fairly regular basis.
But that isn't even speculation. We know the pilots misunderstood what they were seeing, because the telemetry data refuting their misunderstanding is right there on the recording.
To argue their case though, how often does the target tracking system they were using pick up a bird? They may never have seen it do something like that before, and may not even have known it was possible. Also it's one thing for a knowledgeable expert to calmly review and decode the telemetry data at a desk, but the pilot was a little busy also flying an airplane at the time.
If you are talking about navy videos like tictac and gimbal, those videos are probably showing ordinary things through IR camera which makes it look strange. Gimbal videos is even named "gimbal" as the effect seen in the video is cause due to gimbal rotation.
I'm sure US has some kind of secret programme, but I doubt they released the videos showing their own top secret programme. It's much more likely that they released it to throw a bit of a smokescreen and cause their peers like China to expend resources trying to research it. They are releasing some kind of strange patents as well, that seem to be type of pseudoscience which could be a similar play.
When I am watching Battlestar Galactica or most of the other scifi, I am always puzzled to see that the scenarists never take AI to fully operate a spaceship (piloting during landing/takeoff, firing, targeting enemies, etc).
The series starts right after the Cylons, an artificial intelligence created by humans, destroys several planets. Decades before they fought another war when the machines first rebelled. By the time BSG tolls around, any artificial intelligence is taboo and Adama refuses to even enable digital communication between ship subsystems except in one scenario where they had to hard reset all systems to flush a Cylon virus.
to add to this, it's implied (or maybe stated explicitly, I forget) that one of the reasons galactica survived the initial attack is because it's an older warship without the sophisticated computer systems of the newer ships. the newer ships were trivially compromised by the cylons.
> the newer ships were trivially compromised by the cylons.
No.... Caprica six slept with Baltar who gave her access to the defense mainframe. She modified a version of the new fleet control system with the vulnerability that allowed the cylons to take over the fleet.
Galactica survives because she's a museum ship and would never get the update and was built in a time where computer systems were not networked inside battlestars. Another battlestar (spoilers) survives because it was shut down and getting the upgrades.
There are definitely several lines which suggests that Galactica is older than most of the fleet. I recall one line where they say something about how everything is hardwired, and there's no central computer, or something. Something about how it was built during the Cylon wars, which was different than the newer ships which were built after Cylons weren't a threat so they relied on more centralized computer systems.
It's more common in novels and games. But it happens in TV and movies too, take HAL 9000 for example. And you also have to consider that a lot of scifi shows predate deep learning, so it's often referred to as "the computer" instead and AI is reserved for AGI, usually with personality and a humanoid avatar.
In TNG the ship computer handled a lot of routines where the crew only provided high level input. It's advanced enough to run realistic holodeck scenarios.
Stargate Universe also had a ship that was mostly run by the computer and made decisions that were often inscrutable to the occupants and only made sense in retrospect.
As others have noted, BSG specifically has internal story logic which explains why there isn't AI all over the place on the human side. However, more generally, AIs slugging it out, probably over great distances, over timeframes that would probably be very different (both quick and slow) from a human dogfight probably wouldn't be very interesting to viewers relative to essentially WWII dogfights in space.
(And, in general, some version of WWII combat (plus lasers, etc.) is still the model for most SF--even when it's not very deliberately aping it as in the case of Star Wars.
There was a scene in one of the culture novels (I can’t remember now but maybe Excession?) where a ship Mind narrates the progress of a battle to a human onboard.
Forgive my poor memory but it was a really cool battle. The Mind explained methodically and gleefully how it found and destroyed a bunch of enemy ships.
Anyway, at some point the human asks a question and the Mind says something along the lines of, “oh, the battle is already over. It only lasted three seconds”.
Maybe someone can help me remember where this happened.
I believe that was Surface Detail. Having trouble coming up with the exact quote at the moment. I think it was Demeisen of the Falling Outside The Normal Moral Constraints explaining to Lededje.
Long time since I’ve read any of the Culture novels but I’m not surprised there are examples there. I also remember there being scenes in The Forever War. And maybe the expanse although I forget how much of that was AI as opposed to just preparing for high G.
AI's fighting it out while the human crew does nothing is not compelling drama. That's why Star Trek, Star Wars, etc. etc. all have 2nd World War style fighting going on. To get a more realistic idea of what future battles will be like, you need to read Ian M. Banks. In his books, humans are (correctly) too slow and fragile to do any fighting.
Sci-fi is mostly about human entertainment and AI on AI battles just aren't as relatable. Consider something like chess, it used to be quite a drama in the human battles for world champion whereas computer vs computer is a bit meh. The trouble with human vs AI is it tends to be one sided on way or the other.
Actually, drama/soap opera is the part I am not a fan in scifi. I had pain watching Netflix's star trek and see the hero of the saga crying 10 minutes per episode... But I understand, it is here to target a broader audience.
Yeah. I have no idea how a human would compete at all with g-forces alone. Add the fact that you no longer have to design a plane around a human and that definitely gives you a pretty hefty advantage against an f18
Why even put the drone in a fighter? Just put it in a SAM. AI-based guidance system with a multitude of sensors that’s been trained to defeat countermeasures. Seems difficult for a pilot to do much against that.
Perhaps cost? 1 reusable AI platform controlling multiple missiles, vs multiple expendable one. Also I'm guessing maneuverability of a fighter aircraft is much better than a SAM which is more or less ballistic.
Also I'm guessing maneuverability of a fighter aircraft is much better than a SAM which is more or less ballistic.
Not at all. Missiles are far faster and more maneuverable than aircraft. The only chance a fighter plane has against them is with countermeasures such as chaff and flares as well as electronic countermeasures (ECM). All of these are designed to fill the missile’s tracking and guidance systems. Without them, the plane is helpless against the faster and more maneuverable missile.
Missiles are also far cheaper than aircraft. There’s no need to sustain the life and health of a pilot and no need to carry fuel for return trips. This makes missiles very small and light (and thus cheap) compared to a plane.
Planes have another advantage - endurance. Missiles have greater acceleration and maneuverability, but they can't sustain that for as long as a plane can. Which makes "run away" a solution in certain parts of the engagement envelope.
Make a drone that casts away unnecessary parts and turns into a missile when it's close to the enemy.
It can carry half the fuel (or fly twice as long), doesn't need landing mechanisms, can be made with less focus on durability (just needs to survive one flight). It can be very small and fly low (and subsonic) to avoid detection. It could even land and wait near military airports on top of trees etc.
Basically it would work like movable minefields in air. Denying territory to the enemy planes.
Fighters and SAMs serve different roles. A fighter is reusable, can provide coverage at a much greater distance from its launch point and can decide not to shoot/do patrols
SAMs already have as many sensors as can be packed in, and mostly have the human in the loop just to make the decision whether to engage a target. And still cannot reliably get through sophisticated countermeasures.
Airborne platforms for both the big launch sensors and the missile launch point have some inherent advantages is all.
If the plane is high up in the air and moving fast, you have to burn a lot of fuel to make up the difference in energy. Given that they have enough of an energy advantage, they could never get into the engagement range of the surface.
Also, it's really hard to set up a SAM a hundred miles into enemy territory - SAM is fundamentally a defensive technology with extremely limited use for air superiority missions.
Load factor (g force) is actually a small percentage of what makes an effective air to air weapon system such as a fighter jet.
It turns out that the faster, higher aircraft has the advantage. If you have speed, you can deny your better-turning adversary an opportunity to engage you, and you thus dictate the terms of the engagement. If you find yourself on equal footing as a bandit, you can simply climb away at the enemy's maximum level speed and he will be unable to catch you.
Additionally, BVR missiles in particular benefit from a high initial launch velocity so they can use their finite delta-V budget to maximize range and closure with the target.
G Force tolerance only matters if you're in the air combat equivalent of a knife fight in a phone booth.
Methinks these trials coming up will not be fair: a human can continue making good decisions in a non permissive electromagnetic/cyber battle space whereas a drone cannot. Air combat tactics will have to evolve but I don't think we will see drones defeating piloted fighters (in actual combat) in my lifetime. Perhaps if there's a watershed moment in AI but we aren't there yet.
At the ranges you're talking, that's simply too many missiles. To take a SWAG at it: For an aircraft at Mach 1, to hit every potential spot it could occupy in one second from now, you're talking about saturating a 340-metre radius sphere with missiles.
For context, the volume of that sphere is about 164 million cubic metres.
Even if you assume it can't maneuver with complete authority anywhere in that sphere (valid assumption), you could reduce that volume by 75% and still be looking at 40 million cubic metres.
OK, so maybe find out what maneuvers would be the least costly for it, and fire missiles there, and so on down the line, until you run out of missiles. At the very least, it would have to expend tremendous resources to avoid all your missiles.
I mean, you're not trying to blanket the entire space, just put things in its way.
EDIT: Or the missiles themselves can just guide themselves based on tracking how their target moves. Then you won't even need many missiles.
If it's like early AI logic, fly as high as possible, say 40km which turns out is a higher flight ceiling than the attacker in many instances. But the key was the missiles had a ceiling hight of say 20 km, though if launched at 40km, they would still work upon targets flying upwards and the AI logic wouldn't think of them as a threat as those missiles don't work above 20km. Though that means they can climb to that height. Also means if they fire a missile, it won't climb to hit you. Great tactic using the limitations of the missiles to your advantage.
Things like that will be were a pilot will have an early edge, pushing those limits by fully understanding the mechanics of those limits and how they play out. Be that pushing a sonic boom shockwave to effect a small pursing drone. Those for unmanned autonomous system will be the achilles heal in much the same way early Chess AI was able to be beaten by humans thru not doing the obvious most logical move.
But if they want to tune AI for autonomous system, then doing a FTP simulator game, running the NPC drones on a server will get you lots of unique free testing and tuning of that AI done. Be much cheaper and we get a cool game to play.
Maybe, though AI start of somewhere and will it think the missile can fly higher than it thought and change that or learn that launch altitude comes into play. But will always be some hard coded stuff, though the prospect of a drone learning the laws of physics does intrigue me.
I mean, yes. That's how it works. It sounds like you're implying that this is a bad thing, a waste of tax dollars and government spending. I can assure you, there are plenty of other things that the US Gov has wasted money on, probably quite a few more egregious than this.
So even if you believe this was wasted spending, at least that money will hopefully go back directly to the US economy. The military generally prefers to "buy American" when possible, so hopefully at least this tax payer money is flowing back to the country at some level.
I would be less bothered by it if someone was building something other than killing machines. And I really don't care what flag someone waves, nationalism is dumb.
Plenty people build something other than killing machines. And not caring about it is a convenience perspective, afforded by circumstance of living in a well-defended nation.
That's what they want you to think. The war machine is a jobs program, but they could just as easily spend the same money on something more useful like infrastructure projects or social programs.
Well, you're both right. In that the purpose of all this military spending is to make the world safe for Dow, Exxon, and US Fruit, and people who live in the imperial core get cheaper prices because of it.
I guess they'll be very careful with the objective function, not to produce a Kamikaze drone:
- Drone: Less than $10 million.
- Fighter: Almost $100 million and human on board.
It's not far stretched to treat the drone as a new type of missile. Air-to-Air, Air-to-Surface, Surface-to-Surface, Surface-to-Air, and the new generation: Anywhere-to-Anywhere missile.
I don’t see how this will work in a real battle. If the ai is fully self contained with no communication to central command, then it’s like a dumb missile. Once you press fire, you’ve lost control.
If the drone does require connectivity to central command, then shouldn’t it be fairly easy to jam this signal?
Radio spectrum is vast, and you need to focus your jammer energy on the narrow bands used in actual communication. Frequency hopping makes that non-trivial, and that's before you consider directed beams and satellite uplinks.
In addition to frequency hopping, there are spread spectrum techniques to smear the signal across a wider range of frequencies with lower energy in each frequency. Once asked permission to switch on military radio on a project, the answer was “go ahead, we’ll never notice”
That was really just a question of time. Wait a few decades, and see the first warships specialised for carrying drones and missiles (which are essentially kamikaze drones) exclusively.
Well, yeah, you could do vertical landing or water landing. Both have precedent, but water landing probably implies craning the drones on board, which is slow. Vertical landing on the other hand might work well. You'd still need a flight deck for that, but a much smaller one.
I guess something helicopter-carrier sized? I would think the main thing would be to avoid supercarrier sorts of ships. Smaller ships with similar capabilities would mean you'd be able to field more groups.
Launch and landing does not necessarily need to happen in the same platform.
It could be reasonable to have a delivery mechanism for drones that's close to where the action might be (so that the drones can reach the target quickly) but the surviving drones have to go way back somewhere else to land and be refurbished/refueled/rearmed for another mission.
I know you were. What makes a small fighter jet the optimal size for these drones? Or what mission do you imagine for them?
And about the "sizeable contingent". The Ohio class can launch 144 Tomahawks. That is probably enough to suppress the air defences of most countries.
Sorry that I'm picking on these details, it's just people have a tendency to shriek in horror about the future of autonomous weapons, as if we wouldn't be living with them already.
Of course I don't know what the optimal size is. Yet, to be competitive in terms of speed, range, and payload, "small fighter jet" sounds about right to me. At least, if you want to convincingly fill most of the roles that manned aircraft fill in a conventional carrier strike group. That'd include not only things like reconnaissance and weather observation, but also air-to-air and air-to-surface missions over sea and land. So, a kind of hierarchical scheme where the carrier deploys the big drones, which close onto the target before deploying the secondary, disposable missiles/drones. You want to be able to stand off any opposition forces well away from your naval group.
Why specialize them? If SpaceX can land a rocket on a drone-ship, the next step would be to land a drone inside a 20-foot shipping container. Then every container ship becomes an aircraft carrier, capable of carrying thousands of drones. The merchantmen become their own escorts.
Presumably designers are just scratching the surface jettisoning all sorts of current design assumptions that could be removed based on differences in the acceptability of loss, ability to work cooperatively and handle inhuman conditions. You can position the control systems differently, split some of them up, strengthen the airframe by binning almost everything from the cockpit & getting rid of landing gear completely. The gear adds weight and breaks holes in the structure and it's dead weight during most of the key operational tasks - instead use a cooperating "lander" drone to help it land and the risk of that failing isn't as serious without humans. The ability to act with coordination across a group of drones that the article mentions will be hard to beat unless you use a similar array of drones. Seems like the pilot in the plane can't last much longer. Now we just need to see if the budgets involved bring this into being faster than self driving cars!
Re: ditching landing gear, As I understand it, the Gremlins UAV swarm platform uses a ‘live capture’ system from to launch/recover from a C130 - no gear.
You basically turn cargo aircraft into aircraft carriers.
The linked on "parasite fighters" says that they are part of "composite aircraft systems". Before this, I thought the aerospace company Scaled Composites was a reference to materials (like fiberglass etc). The more you know!
The 'Composites' in Scaled Composites is definitely a reference to the materials. From Burt Rutan's wikipedia page:
With his VariEze and Long-EZ designs, which first flew in 1975 and 1979 respectively, Rutan is responsible for helping popularize both the canard configuration and the use of moldless composite construction in the homebuilt aircraft industry.
You might not even need the grappling hook. ‘Just’ stall the drone at a height of a few meters, and have it fall onto a relatively soft surface. Maybe add a small rocket to soften the landing.
> The ability to act with coordination across a group of drones that the article mentions will be hard to beat
Very easy to beat via even somewhat advanced radioelectronic warfare. Can't coordinate if you can't communicate. This is the Achilles heel of all these "advanced weapons systems" that look so great in the powerpoint slides but are only useful against enemies which don't have any weaponry built after mid-80s.
Why are you so confident that communications jamming is easy? The person initiating the communications has all the cards; while the person trying to jam is stuck trying to blast power over a constantly shifting comms frequency and come in via an antenna sidelobe.
I didn't mean it was easy "easy". But it's more than doable. And it's been done in Syria: https://www.thedefensepost.com/2018/05/01/russia-syria-elect.... It's also very asymmetric, similar to e.g. advanced airplanes vs anti-aircraft rockets, or aircraft carriers vs submarines. One of the options is way cheaper, and makes the other not very viable if shit hits the fan.
You speak very confidently but I would like a technical discussion of the points I raised. Even in the link you posted, there is the following: "The U.S. is not getting the message, however, because no one in the U.S. military believes Russia is actually capable of jamming advanced drones and missiles, according to Valeriano."
How do you defeat a drone swarm using frequency agile communications over narrow beams with cheap equipment, such that it's asymmetric in cost?
Well, seems like the jury is still out in many quarters, but there is a secret Cold War going on around electromagnetic jamming tech, and the Russians have the lead:
Of course they'll _say_ they "don't believe" it works. Yet they ceased operations in the region for a few days when Russia turned this gear on. It's propaganda, of course it works. Not as good as the other side claims (that too is propaganda), but well enough to fuck shit up.
I don't know if it's true or if it works, but the article addresses this by saying that they can go into I guess non-networked mode where no one, not even the operators, can contact it. If it gets certain conditions it will send out a message, but otherwise it's that level of autonomy they're assuming.
Yes, but then by definition they aren't communicating, and their effectiveness is massively reduced. They could, for instance, be attacking the same target or even a decoy (wasting ammo), or fly into a trap that other drones would have warned them about. There are also no drones currently that can make firing decisions on their own, thank god, this small scale test notwithstaning. The problems AI has to solve are hard enough even without radioelectronic warfare or adversarial, AI-specific countermeasures. With it they become even less tractable.
I agree that it will be an arms race between AI researchers in competing militaries between ease of getting around defenses and making attacks ineffective.
Of course any country not in on it will be far, far, far, far behind. The idea of a drone swarm based air force going up against a less capable foe seems overwhelmingly superior. I'm sure there will be a give and take as people figure out how to disrupt them better, but I also have no reason to think there won't be ways to compensate for a countermeasure after it's understood.
Or we could just, I dunno, stop having so many wars... but I'm kind of at the point where it's obvious enough that every major world power is working on it that I'm not sure what else the US can do.
Not building $1B fighters seems like a good decision all around if there's any other way... that just seems stupid, no offense intended to anyone working on it but it just seems unreasonable on the face of it if you can make hundreds of unmanned drones instead (or pay teachers more of course...).
I'm curious what will come about with different sizes. I'll bet a lot of people think of this in the context of quadcopters up to predator drones. I'm in the middle of The Diamond Age and all I can think of is the drones so small you can't see them.
Imagine a jet flying through a cloud of drones. Maybe most are destroyed. But some work their way into the internal systems, or perhaps into the cockpit itself, and do who knows what.
I don't think a cloud of drones would be much more effective than the missiles we already have. Unless they are completely undetectable, they would be far easier to avoid than a missile.
Imagine if the goal isn't immediate destruction. Perhaps the jet thinks it destroyed or evaded them. Then it gets back to base.
Drones phone home with location data of base for other ordinance, release biological warfare agents infecting an entire base (or worse), or spread from the infected jet to other vehicles and equipment doing untold billions in damage.
I'm still very sceptical about autonomous drone as fighters.
We already have extremely fast autonomous vehicles specifically designed to take down fighters. They are called missiles. Modern missiles already do everything you are listing including cooperation.
As the main weapons of fighter drones will likely be missiles anyway, the drones will probably be mostly glorified missiles carriers allowing reusability of some of the most expensive avionics.
Usually, the hard problem for AI in military application is respecting the rules of engagement. I fully expect a human to still be in the loop for sometimes.
I think this is going to go like the smartphone, which has assimilated every other consumer electronic device. (music player, camera, video, ebook reader, fitness tracker, etc)
Why not keep drones around for surveillance, for communication, anti-personnel, anti-aircraft, anti-boat, anti-car, signal jammer, take out a cell tower, take out a high tension line, etc...
If you mean having a combination of different specialised drones doing different roles, one of them being carrying sofisticated ordnances to take out fighters then I agree.
If you mean having a multirole drone able to do everything (what would be an AI piloted F25) then I think it doesn't make a lick of sense. Fighters make little sense in modern warfare anyway. The cost/capability ratio just doesn't work especially considering they are only ever going to be used against considerably less advanced opponents as all the nations operating modern fighters could nuke each other.
Regarding AI, my personnal belief is that having a computer assisting a human piloting a group of drones is a more promising avenue than fully autonomous drones but I might be wrong.
> If you mean having a combination of different specialised drones doing different roles, one of them being carrying sofisticated ordnances to take out fighters then I agree.
That is the opposite from what I'm saying.
For example, all the current F planes are not specialized -- they are all termed multi-role fighters. They have all been treated as a platform and upgraded for more capabilities over time.
I think drones will follow the same path -- glomming on capabilities instead of becoming more specialized.
I have to wonder when it will switch to small, inexpensive drones instead of the larger, jet-type ones. Imagine fighting against a swarm of 10,000 drones (each with a small explosive, EMP, etc). Similar to the Fire Bats[0] in WWII.
Never. Those things will get burnt by laser weapons if they are too small to be efficiently targeted by cannons and missiles. They also won't have range, payload for missiles with range, etc.
They can be useful for other roles. Surveillance, ECM, flying in formation and coordinating to simulate a larger aircraft, tricking passive radar sites to go active, etc.
Mainly these two out of your list - surveillance and tricking passive radar sites to go active. The former is being done today, the latter is being solved by loitering munitions that hang around until a site does go active.
> What’s the difference between a Tomahawk cruise missile and a drone?
With human pilots, there is a bright line separating aircraft and missiles (although even that line wasn't completely firm - see Kamikazi attacks). Taking the pilot out of the equation, that line dissolves into a gradient.
> Is the distinction that drone can return to base, or otherwise land and be reused?
A Tomahawk is a drone, just not the kind discussed in this article. A Tomahawk is subsonic missile designed to attack land and sea targets. The drone they are talking about here would be able to attack land, sea, and air targets. It would be vastly more maneuverable, faster, reusable, have AI specific to air-to-air combat, and I assume it would have greater versatility in selection of its munitions during an engagement.
Is this incredibly obvious or did I miss the part where they specified the rules of the game? I assume they're not actually going to have the drone fire actual missiles at the manned fighter...right?
I would expect them to use statistical models to estimate probability of kill given the state of the target, drone, and environment at the time of a simulated weapon firing. Certainly not perfect, but you probably won't find many test pilots willing to have a live missile fired at them. Instructors are able to judge pilot effectiveness from simulated firings, so it should be good enough to evaluate a drone.
It will be great if in the future war is simply a function of how much money you’re willing to spend to destroy a target. Probably not much anyone can do to defend against a swarm of drones that never gets tired, never misses targets and doesn’t even care about death.
193 comments
[ 3.3 ms ] story [ 253 ms ] threadWe can do this all day but I don't think it's helping anything.
If it only has to "live" for a few minutes, you can get a LOT more performance out of it, which is exactly what they're designed around.
But it's a trade-off. You could invest in a (practically) infinite supply of global capable missiles to be launched from the literal center of America, and to get 24 hour coverage you just launch one every 30 minutes. But that's extremely expensive and stupid.
But a missile which is carried to be closer to the battle can be smaller, and cheaper. Then you need some kind of vehicle to get it there. This kind of thing is that vehicle.
https://en.wikipedia.org/wiki/Millennium_Challenge_2002
Finally the fly boys will realize they're totally outmatched and that the F35 will be our last fighter. In my wildest dreams it gets cancelled and a bunch of smaller defense companies spring up to bring prices way down and competition in the space increases. Then overall defense spending goes down and we get more money for services - while being still being armed to the absolute teeth.
It looks like the drone in question here is the following (quoted from Aerospace Testing International), also mentioned in the linked source:
'The “fighter-sized” 5th Generation Aerial Target (5GAT) is 12.2m (40ft) long, a 7.3m wingspan and a maximum gross weight of 4,350kg (9,600lb). It is designed to be launched and landed using a conventional runway. The drone features two afterburning jet engines and a 95% carbon fiber airframe.' It seems to be designed specifically to stress test our own flights as target practice and doesn't seem like it's actually going to be going into combat anytime soon.
https://en.m.wikipedia.org/wiki/Macross_Plus
https://en.wikipedia.org/wiki/Northrop_YF-23
Macross Plus being much older doesn't attempt to get into AI for the opponent, and instead uses something a little more down to earth at the time which was neural interfaces.
That's a statement nobody here will have problems to accept right? I expect nobody can think of any example that makes "more accurately" a problematic claim.
Combat is a simpler zero sum game, pick and target and destroy.
Based on how AI plays top games right now(Starcraft and Dota), very fast quick reflexes and pin point accuracy of their attacks I think the AI drones are going to crush humans.
https://www.youtube.com/watch?v=mfhAC2YiYHs&t=986s
But that isn't even speculation. We know the pilots misunderstood what they were seeing, because the telemetry data refuting their misunderstanding is right there on the recording.
To argue their case though, how often does the target tracking system they were using pick up a bird? They may never have seen it do something like that before, and may not even have known it was possible. Also it's one thing for a knowledgeable expert to calmly review and decode the telemetry data at a desk, but the pilot was a little busy also flying an airplane at the time.
I'm sure US has some kind of secret programme, but I doubt they released the videos showing their own top secret programme. It's much more likely that they released it to throw a bit of a smokescreen and cause their peers like China to expend resources trying to research it. They are releasing some kind of strange patents as well, that seem to be type of pseudoscience which could be a similar play.
No.... Caprica six slept with Baltar who gave her access to the defense mainframe. She modified a version of the new fleet control system with the vulnerability that allowed the cylons to take over the fleet.
Galactica survives because she's a museum ship and would never get the update and was built in a time where computer systems were not networked inside battlestars. Another battlestar (spoilers) survives because it was shut down and getting the upgrades.
You can both be right.
I forget if there was anything special with the computers at that point
(And, in general, some version of WWII combat (plus lasers, etc.) is still the model for most SF--even when it's not very deliberately aping it as in the case of Star Wars.
Forgive my poor memory but it was a really cool battle. The Mind explained methodically and gleefully how it found and destroyed a bunch of enemy ships.
Anyway, at some point the human asks a question and the Mind says something along the lines of, “oh, the battle is already over. It only lasted three seconds”.
Maybe someone can help me remember where this happened.
drone could have higher max speeds, higher max acceleration. fighter was designed ... what 20 years ago?
Cost of drone? Cost of loss of either? total cost of ownership is vastly different.
There is real cause to be very afraid.
Not at all. Missiles are far faster and more maneuverable than aircraft. The only chance a fighter plane has against them is with countermeasures such as chaff and flares as well as electronic countermeasures (ECM). All of these are designed to fill the missile’s tracking and guidance systems. Without them, the plane is helpless against the faster and more maneuverable missile.
Missiles are also far cheaper than aircraft. There’s no need to sustain the life and health of a pilot and no need to carry fuel for return trips. This makes missiles very small and light (and thus cheap) compared to a plane.
It can carry half the fuel (or fly twice as long), doesn't need landing mechanisms, can be made with less focus on durability (just needs to survive one flight). It can be very small and fly low (and subsonic) to avoid detection. It could even land and wait near military airports on top of trees etc.
Basically it would work like movable minefields in air. Denying territory to the enemy planes.
Airborne platforms for both the big launch sensors and the missile launch point have some inherent advantages is all.
Also, it's really hard to set up a SAM a hundred miles into enemy territory - SAM is fundamentally a defensive technology with extremely limited use for air superiority missions.
A thousand years for now, it may be up to Brion Brandd to save this planet... https://www.goodreads.com/review/show/2559973349
It turns out that the faster, higher aircraft has the advantage. If you have speed, you can deny your better-turning adversary an opportunity to engage you, and you thus dictate the terms of the engagement. If you find yourself on equal footing as a bandit, you can simply climb away at the enemy's maximum level speed and he will be unable to catch you.
Additionally, BVR missiles in particular benefit from a high initial launch velocity so they can use their finite delta-V budget to maximize range and closure with the target.
G Force tolerance only matters if you're in the air combat equivalent of a knife fight in a phone booth.
Methinks these trials coming up will not be fair: a human can continue making good decisions in a non permissive electromagnetic/cyber battle space whereas a drone cannot. Air combat tactics will have to evolve but I don't think we will see drones defeating piloted fighters (in actual combat) in my lifetime. Perhaps if there's a watershed moment in AI but we aren't there yet.
Why isn’t auto-aim game over here for both sides?
Approach and simply fire missiles at all the spots the plane could be in the next few seconds. Done.
For context, the volume of that sphere is about 164 million cubic metres.
Even if you assume it can't maneuver with complete authority anywhere in that sphere (valid assumption), you could reduce that volume by 75% and still be looking at 40 million cubic metres.
I mean, you're not trying to blanket the entire space, just put things in its way.
EDIT: Or the missiles themselves can just guide themselves based on tracking how their target moves. Then you won't even need many missiles.
That's what many types do nowadays.
Things like that will be were a pilot will have an early edge, pushing those limits by fully understanding the mechanics of those limits and how they play out. Be that pushing a sonic boom shockwave to effect a small pursing drone. Those for unmanned autonomous system will be the achilles heal in much the same way early Chess AI was able to be beaten by humans thru not doing the obvious most logical move.
But if they want to tune AI for autonomous system, then doing a FTP simulator game, running the NPC drones on a server will get you lots of unique free testing and tuning of that AI done. Be much cheaper and we get a cool game to play.
So even if you believe this was wasted spending, at least that money will hopefully go back directly to the US economy. The military generally prefers to "buy American" when possible, so hopefully at least this tax payer money is flowing back to the country at some level.
I hope for Maverick's sake aerial combat and chess are two different things.
- Drone: Less than $10 million.
- Fighter: Almost $100 million and human on board.
It's not far stretched to treat the drone as a new type of missile. Air-to-Air, Air-to-Surface, Surface-to-Surface, Surface-to-Air, and the new generation: Anywhere-to-Anywhere missile.
If the drone does require connectivity to central command, then shouldn’t it be fairly easy to jam this signal?
It could be reasonable to have a delivery mechanism for drones that's close to where the action might be (so that the drones can reach the target quickly) but the surviving drones have to go way back somewhere else to land and be refurbished/refueled/rearmed for another mission.
And about the "sizeable contingent". The Ohio class can launch 144 Tomahawks. That is probably enough to suppress the air defences of most countries.
Sorry that I'm picking on these details, it's just people have a tendency to shriek in horror about the future of autonomous weapons, as if we wouldn't be living with them already.
You basically turn cargo aircraft into aircraft carriers.
https://www.darpa.mil/program/gremlins
https://i.redd.it/4deoeuq9e4m21.png
Drones are the game's main theme and it's great.
https://www.youtube.com/watch?v=8mkmBWIUT54
With his VariEze and Long-EZ designs, which first flew in 1975 and 1979 respectively, Rutan is responsible for helping popularize both the canard configuration and the use of moldless composite construction in the homebuilt aircraft industry.
https://en.wikipedia.org/wiki/McDonnell_XF-85_Goblin
See https://m.youtube.com/watch?v=_s_Q5CI7p5M and https://m.youtube.com/watch?v=f7Lu6LEQ0zo for inspiration. I doubt those landings were healthy for the pilot’s back, but for drones, and given half a century or so of technological development, this might work.
You might not even need the grappling hook. ‘Just’ stall the drone at a height of a few meters, and have it fall onto a relatively soft surface. Maybe add a small rocket to soften the landing.
Very easy to beat via even somewhat advanced radioelectronic warfare. Can't coordinate if you can't communicate. This is the Achilles heel of all these "advanced weapons systems" that look so great in the powerpoint slides but are only useful against enemies which don't have any weaponry built after mid-80s.
How do you defeat a drone swarm using frequency agile communications over narrow beams with cheap equipment, such that it's asymmetric in cost?
https://www.voanews.com/europe/russian-mystery-weapon-claim-...
Of course any country not in on it will be far, far, far, far behind. The idea of a drone swarm based air force going up against a less capable foe seems overwhelmingly superior. I'm sure there will be a give and take as people figure out how to disrupt them better, but I also have no reason to think there won't be ways to compensate for a countermeasure after it's understood.
Or we could just, I dunno, stop having so many wars... but I'm kind of at the point where it's obvious enough that every major world power is working on it that I'm not sure what else the US can do.
Not building $1B fighters seems like a good decision all around if there's any other way... that just seems stupid, no offense intended to anyone working on it but it just seems unreasonable on the face of it if you can make hundreds of unmanned drones instead (or pay teachers more of course...).
Imagine a jet flying through a cloud of drones. Maybe most are destroyed. But some work their way into the internal systems, or perhaps into the cockpit itself, and do who knows what.
Also, poor animal.
Drones phone home with location data of base for other ordinance, release biological warfare agents infecting an entire base (or worse), or spread from the infected jet to other vehicles and equipment doing untold billions in damage.
We already have extremely fast autonomous vehicles specifically designed to take down fighters. They are called missiles. Modern missiles already do everything you are listing including cooperation.
As the main weapons of fighter drones will likely be missiles anyway, the drones will probably be mostly glorified missiles carriers allowing reusability of some of the most expensive avionics.
Usually, the hard problem for AI in military application is respecting the rules of engagement. I fully expect a human to still be in the loop for sometimes.
Why not keep drones around for surveillance, for communication, anti-personnel, anti-aircraft, anti-boat, anti-car, signal jammer, take out a cell tower, take out a high tension line, etc...
oh yeah, add contact tracing with an update :)
If you mean having a combination of different specialised drones doing different roles, one of them being carrying sofisticated ordnances to take out fighters then I agree.
If you mean having a multirole drone able to do everything (what would be an AI piloted F25) then I think it doesn't make a lick of sense. Fighters make little sense in modern warfare anyway. The cost/capability ratio just doesn't work especially considering they are only ever going to be used against considerably less advanced opponents as all the nations operating modern fighters could nuke each other.
Regarding AI, my personnal belief is that having a computer assisting a human piloting a group of drones is a more promising avenue than fully autonomous drones but I might be wrong.
That is the opposite from what I'm saying.
For example, all the current F planes are not specialized -- they are all termed multi-role fighters. They have all been treated as a platform and upgraded for more capabilities over time.
I think drones will follow the same path -- glomming on capabilities instead of becoming more specialized.
We will have multi-role drones.
[0] https://www.wikiwand.com/en/Bat_bomb
https://news.ycombinator.com/item?id=23423240
There are examples in media!
Many mechas from the Gundam series use funnels, a form of drone, in order to overwhelm a single enemy or attack multiple ones:
https://youtu.be/qc6ME9S_qps?t=1m5s
Ace Combat 7 features a massive battle against an airborne UAV carrier and its swarm of drones:
https://youtu.be/JQJpRwkXHj4?t=19m50s
Sometimes the entire screen is filled with targets!
Fiction: https://m.youtube.com/watch?v=9CO6M2HsoIA
Reality:
https://m.youtube.com/watch?v=ndFKUKHfuM0
https://m.youtube.com/watch?v=BF1sJjk66rU
https://www.theverge.com/2019/8/9/20799148/darpa-drones-robo...
https://www.usatoday.com/story/nation/2014/11/10/aircraft-ba...
"The Alarming Case of the USAF’s Mysteriously Missing Unmanned Combat Air Vehicles"
https://www.thedrive.com/the-war-zone/3889/the-alarming-case...
Tomahawk is already (2015) available with reconnaissance camera and loiter mode.
Is the distinction that drone can return to base, or otherwise land and be reused?
I suspect 'drone' is a catch all term for things that don't have more specific names.
With human pilots, there is a bright line separating aircraft and missiles (although even that line wasn't completely firm - see Kamikazi attacks). Taking the pilot out of the equation, that line dissolves into a gradient.
> Is the distinction that drone can return to base, or otherwise land and be reused?
That's probably the most useful distinction.
Further, the drone may not necessarily used for combat, it could be used as a decoy.