F-22

last SG · 2003-04-28T19:35:15+00:00

scully next time swarm with UAV, then got a few elite manned pilots shoot down many UAV at one shot… wha like science fiction story like that…..

SingaporeTyrannosaur SG

Don't be too surprised, but that's exactly the case! The US military is actually calling on a team made out of game designers, military people, and DoD people to come up with mission preformance AI for their next generation of advanced cruise missiles and UCAV platforms. And that includes AI that will enable such platforms to actually tangle in active combat, in addition to evasive and mundane mission subroutines.

What is happening is that the idea is to make the AI itself modular and seperate from the rest of the software that actually does the targeting, flying the aircraft and what so. And that's what the US military is doing now, calling on game developers to help them in the development of their artifical intelligence.

I also remember a slight test they did in the early 1990s in a book I read some time back that I am now trying to find, where the put a few simple automated hardcoded subroutines into a QF target drone (the one they usually use to haul radar reflectors to confuse enemy radar in one of shunk works stealth programs). These commands told the drone how to react whenever a mounted threat reciever told them that they were being targeted or there was something chasing them, and basically that was to go into a series of erratic crazy 20 gee manuvers and pump out chaff and flares.

And they launched that thing from a B-52, and pitted it against F-15 aircraft, which were ordered to try to shoot it down. The drone survived 50 dogfights and dodge no less then 24 aircraft and many times that number of simulated missile launches against it.

I believe that was a study into the feasibility of making drone aircraft and cruise missiles smart enough to avoid getting shot down, but I think the interest in such research was dropped in the favour of making more smart munitions and it was not remembered until recently.

And all this was done with computing technology at the time when a good flight simulator was considered Microprose's F-15 Strike Eagle 2.

I'm still trying my best to dig out the article so I can quote it, but I remember it quite clearly. I suspect it could be buried in one of the 12,000 pages of TakeOff Magazine I have.

The whole gist of the matter is, if anything this experiment proves, you dun need super crays to make computing power for a UCAV to fight fighter pilots. If you write good enough software for a automated UCAV to dodge and run through a evasive routine whenever it determines it is in threat. And also subroutines to make it aquire and fire back weapons at that theat, and do all that stuff while following a mission flight plan. There is no reason to see why UCAVs cannot hold their own against manned aircraft.

And better, since UCAV can be designed to make turns in several times multitude that of a normal fighter pilot can, it will be no trouble for the UCAV to just snap about a J turn to face it's opponent (assuming it even needs to face its opponent, given development into networked sensor enhanced AA missiles) and snap off a shot. It does not have to engage in a game of energy manuvering in a limited airframe playing airborne chess until somebody screws up. There's no need to, when you can turn far tighter then your opponent.

I think the logic will be self-apparent if you ask fighter pilots if they want to dogfight something that can make 30 gee turns and snap around to face them (if you calculate, that's a 180 turn taking only 1 second at 450 knots while a manned aircraft takes 3-4 seconds to make the same reversal at nearing 10 gees) assuming they can even get on it's tail.

In the light of this, I doubt there will be much dogfighting in a sense against UCAVs at all, not unless manned aircraft can keep up and preform UCAV like manuvers. No matter how good the pilot, the plane also plays a very big part. The red baron in his triplane is not going to prevail against an average fighter jock F-16 no matter how many times better the german may be, not unless the F-16 pilot does something stupid that is.

SingaporeTyrannosaur SG

But these are not intracatable problems, like that of the physical limits of pilots. You are assuming that there can be no safe system enough for the UCAV to operate (in that case, the USAF might as well scrape them right now). Most of these concerns you stated of system security seem to be unfounded, partly so because these are part and parcel of any military operation. And obviously the USAF has figured out these problems to a large enough extent that their long term plan (sorry, not near moonbase era) is to replace a large portion of their force with unmanned platforms. You are talking about the replacement of many makes of manned aircraft like the F-16, A-10 and what so have you with something entirely new and different.

What is evident from the evidence we are seeing is that, UCAV technology has advanced to a point where it is possible to operate them, and do so combat effectively. What it indicates, is that these vulnerabilities can be, and have been managed. It is certainly not foolproof, but apparently reliable enough that the USAF is hardly expressing pessimism about such technology. Not unless you want to take the fool's path of arguing for 100 percent security that is, which is impossible.

On the other hand, history has shown manned aircraft are far less reliable in terms of survability and durability and increasingly expensive.

440mhz SG

I agree with ST in the area of manned aircraft vs. UCAV dogfighting. The rules here are rewritten.

That is because funky (and basic) air combat moves like high yoyo, low yoyo, esp the sis and the verti version which are staple moves of air combat only applies to aircraft that are more or less similar in preformance envelopes, with turn factor being the king factor.

That's why countries spend so much to try to be the top of this by making aircraft that can do these kinds of moves the best. If you can't do all these air combat moves the best, that means to have the most energy and turn the tighest (the two main things of air combat school) are those who will fly home from the fight.

The problem is, humans cannot take accelerations above 10 gs, and anything at 13 or more sustained will kill. So this means that the best yo yo or vert. sis can only be done at most 10 (even this is not adviced, because most pilots will be near enough to g-loc that their SA is zero and they are sitting ducks)

Hence we see more creative combat moves like J-turn, extreme AoA and cobra coming up these days that are trying to afford solutions to get the all important nose around without killing the pilot. But these are only stopgap measures.

The UCAV rewrites all these rules. Without the pilot, the designers can make an airplane that can easily pull over 30 g and even do negative g moves with the same amount of force (humans die at negative 5 g)

This completely rewrites the air combat school book. If a pilot is pitted in a dogfight against a UCAV, he will immediately lose because he simply can't outmanuver the UCAV. In fact, the UCAV does not even have to be caught up in all the staple air combat moves that fighter school teaches because it is too far ahead of the fighter pilot abilities to even be caught up in anything like a vert. sis fight (something we always see in the movies but very rare and only happens when true aces meet, like Cunniham and Col. Tomb in vietnam).

ST is correct to say that they dun need to make the UCAV fight the fighter pilot on his terms and turf, that's because when they meet, the fighter pilot will have to try to outfly the UCAV on it's terms. And that's going to be very difficult, meaning the pilot is at a disadvantage.

|-|05| SG

I think the logic will be self-apparent if you ask fighter pilots if they want to dogfight something that can make 30 gee turns and snap around to face them (if you calculate, that's a 180 turn taking only 1 second at 450 knots while a manned aircraft takes 3-4 seconds to make the same reversal at nearing 10 gees) assuming they can even get on it's tail.

i know this is just an example but the drone will stall.As for the rest of your arguement it makes sense accept that the technology to make the drone suppass the human pilot is not yet here.It's getting there but not there yet.For now the technology suits a strike mission and as a recon mission.Why?Strike missions and recon mission are "easier" due to the target package being fixed meaning less programming is required.However for fighters and air support roles this is different.In addition to needing to put in computers to handle the data from the human pilot it has to put in computers that can make decisions on it's on should the signal be jammed.That means putting multiple situations and how to handle it.The computer that can make the millions(maybe billions) of calculations required is currently way too big.Which is where the human is better...the brain is smaller and capable of learning from mistakes which the computer cannot YET.Sure you can take the AI from a computer game and put it into a drone but really if the Ai from that game could be beaten by some1 like us how u'd expect it to do againest a real pilot??And from the programming of most AI's currently(namely Janes and Falcon 4.0)the game is programmed to respond on act to your moves.It does things on a 1 to 1 basis and is not yet able to handle the calculations at the speed needed to be fully independent.Now they need at least 2 computers being able to do the same job on each drone to make it more reliable and less likely to fail.Meaning if each computer is bigger then a PC it's too big to be placed in each drone.For now the technology is still too experimental to be used like the stealth fighter in the 70's.Give it another 10-15 years to grow and mature.

And exco nothing is 100% reliable but for now they level of relibility of a computer is not yet there.Imagine your home PC crashes often enough and it is not even doing alot of work and those computers that do not fail easily are the size of cars!So now it is not 100% nor will it ever be.But it can still be used however not just yet!

SingaporeTyrannosaur SG

Bang on the spot 440 dude.

EXOC's points on dogfightign fall apart when you start to take it down to the base level. Forget all the humanistic rethoric about dogfighting being dribbled out as if gospel. Let's go to the basics of dogfighting, what it really is about.

The ability to do moves like 30 gee turns and that much minus gees opens up a whole regime of short-cut air combat that humans can never reach.

This is very important, considering that most of the air combat moves you mentioned are tailored to human and manned aircraft limits. While the pilot has to take the long way around in a vertical scissors to get his nose from A to B because he cannot take minus g, the UCAV can just snap about into a -30 turn and be on him. Heck, the UCAV does not even need to be involved in the vertical sissors to get its robotic sights on the pilot's butt! Boom!

You are right to say aircombat between the two will be decided on the UCAV's terms, not the manned aircraft. The UCAV does not need to outthink the human, it just needs to outfly him. You can have the darndest dogfight idea in the world but it's irrevelant when the UCAV beats you to the punch and gets on your tail before you can implement it in response to a move which was already long gone.

Say, how do you know so much about dogfighting dude?

Oh yah |0-5|, I think you should know that most planned UCAVs of the future will probably be no smaller in size then modern combat aircraft for all the systems, minus 30 percent weight, cost and size for life support and human interface systems. I'm sure they could fit a few computers in, and not typical PCs, but very high end units that are certainly much smaller and far more powerful that you can fit into a black box. To say that UCAVs can never fight fighter pilots may be as myopic to see Mercury, and claim that Apollo is not possible.

The idea is, the tech is already present, and it is growing.

SingaporeTyrannosaur SG

What do you mean? Shunk Works has already made drones for some time that can snap in fifty, yes, 50 gee turns as if it was a sunday afternoon tea. Those things turn 5 times faster then the F-22.

The reason why I decided to stick to 30 gees is because it was the figure the USAF military planners stated and since UCAVs are larger then the Shunk Work drones.

|-|05| SG

It's precisly that the plane will be smaller that a present day computer cannot be used.Top of the line computers are not getting small they are getting bigger infact.The computers that are used on board ships to automantically fire the air defences are already small compared to super computers and those are already something like 4m by 4m.No computers are not yet there.They currently suck up too much power,are too big and not yet that reliable.Basically you take the B-2's computer multiply it by 3-4 times and try to fit that into an f-16 sized plane.Too big!!And i did not say never i said NOT YET.Technology is still to experimental and not yet reliable.However when it does come it will be better in some ways then a human due to the aiblity to make more calculations per sec

|-|05| SG

I'm not that erm.....that knowledgeable in that area(the ability to turn 50g's) but i still think that making sure a small turn with that small turn radius is dangerous to the situation due to the lose of speed.However the ability to make bigger negative turns is much more erm......brighter?

|-|05| SG

oh and 1 more thing there will not need for a full 360 cam or even a 180 cam.Just a wide angle cam of say more then 120 deg is enough.Couple that with a VR helmet should do it.

MushyMaro SG

i read the point on the "human" controller on the ground transmitting instructions for the UCAV? well that is a no no...
the lines of transmission is not possible (many factors) and it is useless... the idea of having UCAV is to have IA control and please don't underestimate the power of IA... they ARE written by HUMANS...

Joe Black SG

I have been reading this thread with great interest, and here is my 2 cents worth:

I believe from all the reading so far that UCAVs for now and in the very near future, would not be able to replace or defeat a manned agile aircraft like the Typhoons, Raptors, SU-30 series, Rafales, etc in an air to air combat. Reason, simple enough, UCAVs cannot think tactics. Yes, as SgDino has pointed out that you can make a UCAVs go zip-zap to avoid incoming missiles, pulling high Gs. Yes, a UCAVs will be able to pull GT 9gs whereas a human pilot is limited to not more than 9. However, today's AAMs can certainly pull higher Gs than any UCAVs. Rafael Python 4 can pull 60gs. AIM-9X can do similar. The no-escape zone for AAM will encompass any UCAVs anyone can built. This essentially negates a UCAVs advantage to pull higher Gs than human aircraft theory. It is the first look kills scenarios. In this first-look kill scenarios, I have to say that a skilled human pilot with a 20-20 vision will win the day. Yes, I have no doubt that UCAV kill ratio will be close, but I think so far technology cannot substitute human's adaption and being able to react faster to tactical changes, ie. humans are capable of individual thinking whereas machines are being taught. Even current and near future AI wouldn't come close to human's ability. This is the very same argument used in Star Trek Next generation's Data character.

Secondly, most future A2A battle will not be fought in WVR environment. It would naturally be BVR environment. And logic tells us that the bigger the platform, the bigger the radar the platform can carry. The bigger the radar, the longer distant it can detect. Thus once again, manned aircrafts win.

Ok, lets talk about current UCAVs development. Taking from 2 different streams of development, eg, US UCAVs, and Sweden UCAVs.

US UCAVs are basically designed for ground strike modes. That calls for the UCAVs to be design with high degree of stealth, range and payload capabilities. The UCAVs can be pre-programmed, with some limited remote control from the ground or some large airborne platforms. It is almost like a smart and agile Global hawk cross Predator kind of thing.

Swedish UCAVs are basically controlled by a manned 2 crews Gripen. Designed primary for ground attacks too, with a high degree of stealth.

Note that both UCAVs are designed not to fight air superiority fighters but use for ground attacks. Their roles confined to limited human losses in strike roles. If UCAVs were available in the first gulf war, then the allied air losses would be cut down by some 50 to 90%.

What I have stated here could be located all over the web and I am simply too lazy (right now to find the references) Smile

My conclusion is that for the next couple of decades, we would see UCAVs replacing manned roles but will NEVER totally replace manned aircraft - especially in interception, escort, air-superiority and supremacy roles. If UCAVs can defeat human piloted aircrafts in A2A engagement, God bid and help us, we will really be facing the Matrix/Terminator scenarios. Human race will be made irrelavant.

EXCO SG

This is exactly my point. It can be used, not just yet! When? Thats the secondary issue here.

I understand SingaporeTyrannosaur's enthusiasm and knowledge about unmanned technology, however it is not as simple as AI for a game. You have told me that it is, and you just turned me off from this discussion. I don't what led you to thinking that unmanned technology has a link to do with game AI, perhaps on a very abstract level.

Also, do not forget that currently, we are talking about UAVs which do not operate in force. Once again, what is being suggested is that a whole air force of UCAVs. Herein lies my point about networking. How do you, or can you coordinate perhaps 50 of them all at once? Imagine the resources, and the computing power needed at home base.

Dude, what made you think that dogfighting is simply about pointing in the correct direction and shooting?

SingaporeTyrannosaur SG

Secondly, most future A2A battle will not be fought in WVR environment. It would naturally be BVR environment. And logic tells us that the bigger the platform, the bigger the radar the platform can carry. The bigger the radar, the longer distant it can detect. Thus once again, manned aircrafts win....

My conclusion is that for the next couple of decades, we would see UCAVs replacing manned roles but will NEVER totally replace manned aircraft - especially in interception, escort, air-superiority and supremacy roles. If UCAVs can defeat human piloted aircrafts in A2A engagement, God bid and help us, we will really be facing the Matrix/Terminator scenarios. Human race will be made irrelavant.

[/b]

Erm actually military analysts indicate that in a BVR engagement, a UCAV has the inherent advantage. This is because the UCAV will inherently be able to carry the critical 30 percent more sensors or weapons or anything else you want to cram in the platform then a manned aircraft. You point only works when you suggest that UCAVs will be inherently smaller then that of modern combat aircraft, which does not have to be always true. Even going by your reasoning, it can be reasoned that a UCAV with similar radar capability will be 30 percent smaller, and also we are counting in the stealthier nature of the UCAV because it's form does not have to accomidate a human pilot (50 percent of the F-22 rcs comes from it's canopy). In a BVR, a UCAV will have a large advantage.

As for no-escape zones, you must be aware that it is not fixed and differs with aircraft with difference preformances. Usually the figure you see for no-escape zones for most missiles you read about have been tuned to a maximum of the 9-10 gees target manuvering envelope. The tighter the target can turn (or faster it can fly), the smaller the no-escape zone becomes. If you use an AIM-120 to try to shoot down an aircraft, the no-escape zone is certainly much larger then if you are trying down to shoot down, say another missile.

Hence a manuvering UCAV's no escape zone is actually three times smaller, and manuvering aside, a computerized UCAV can react and pump out countermeasures with timing far better then that of any human pilot before, or after missile launch. Hence the chances of it coming out unscathed is far in excess of that of a manned aircraft, as a simple QF drone has already shown. To get within a UCAV's no-escape zone for your missile, you have to manuver into a envelope three times smaller, and I doubt these things present themselves in air combat much.

I'm not sure how getting show down by robots makes humans irrevelant. These things were made by humans in the first place. Besides, robots have been shooting down humans in air combat for a long time liao, witness all the air to air kills with missiles Wink

SingaporeTyrannosaur SG

On networking:

Actually in terms of networking, I think you might have to upgrade your knowledge and figures by a large number of manitudes. Already the US army has experimented with networking up to 2,000 micro drone buggies that they intend to launch from a mothership to overwhelm a house with tasers or tear gas as their subsitute for bloody FIBUA. It's one of their unmanned projects and I'll pass you the link once I can scrounge it up. The fact is, such networking is far from impossible.

To understand how, what you are suggesting is a top down network that has to share all the resources in between. Such a network is based on a model of how human organizations usually do things.

But the network model that has come up with to figure out how to network multiple units with little latency is the bottom up, or ant-colony approach. Where units are networked together in a form of collective intelligence, rather then centralized intelligence. This delocalized network model makes it possible to link up an indefinate number of units, coordinate and make them them do the stuff you want as a team.

The ant-colony approach that they are now look to. This also means that the network is incredibly robust, because it's intergity is decentralized. The only way to destroy this network is to kill every single one of the drones. Such a network requires minimal external computing power to support at all as the drones all provide to the overall coordination and network, like an ant colony. As a matter of fact, the more drones you have the better, because the overall oordination and ability of the network will increase. Of course, the disadvantage to this system is that you have less localized control over each individual, as compared to the computing extensive traditional model you suggested. But this is the network model defence planners intend to use for their future extensive drone networks, as it applies to the supernet of other sensors and units.

On dogfighting, I did not come up with that idea. One F-15 pilot said it pretty much himself during aerospace to sum up dogfighting.

"It's nothing like the stuff you see in the movies that you have to come up with complex plans to outfly your enemy. It's just basically a matter of pointing you aircraft in the right direction faster then he can, and shooting."

When asked on what that meant, it was that dogfighting was a brutally simple affair that usually only takes snap seconds to conclude, tatical thinking and human brain mumbo jumbo greatly absent. Get you nose right fast and blast him out before he can.

SingaporeTyrannosaur SG

It's that case in traditional airfoils. But modern airfoils and exotic aircraft forms can easily make snap turns, go into controled stalls and reverse heading. The problem is, while all these exotic manuvers have been made possible with emerging technology, the basic limit of human endurance is still there. Humans can only do these manuvers under limited circumstances lest they kill themselves if they push it too far. If you do a J-Turn too hard, which involves negative as well as side to side gees (usually takes about 8 seconds to do) you'll stroke yourself and turn your plane into a flying coffin that will intersect the ground shortly. A UCAV can exceute the same manuver easily eight times faster.

A UCAV can take full advantage of this new genre of flying, and yes, they provide an extreme advantage in combat. To fight a UCAV in air combat would be a nightmare, because it will be akin to flying against a airborne turret that can face anywhere it wants far faster then you can. Hardly anything I would go into with the pretext that the human will always prevail.

If the human could fight the UCAV on fair terms and even ground, I'm sure we'll win more of the times, but this is not the case. The UCAV has inherent advantages in air combat manuvering, and these advantages are huge, akin to the difference between a camel sopswhip and a F-8 Crusader. It's a revolution of difference.

EXCO SG

I would like to see the source on that. If you don't mind can you post the url?

What I am thinking is that the "2000 micro drone buggies" is nowhere near the sohistication of UCAVs. Like I mentioned earlier, the data being transferred to and from UCAVs is not as simple as other UAVs. You have real-time graphics, radar, telemetry data, weapon datalinks and whatnot. The example you have mentioned is a large narrowband network at best. What the reality about is is a huge broadband network.

One F-15 pilot? He must be in publicity mode. And if he said this - "tatical thinking and human brain mumbo jumbo greatly absent", I seriously doubt your source.

MushyMaro SG

lets us see in the next 10 years progress... if the IA controlled plane can just land itself without any support from the ground.... like ILS... and others. if so ... i think pilots are already halfway out of the ring...

take a bet.
i bet on the UCAV...only if there is no powerful electronic Area effect weapons around.

|-|05| SG

The ant colony idea is actually what makes the drones simply that drones and not AI.Ant colony works on the principal that each part of the network is feeding off each other.Like a human body.1 part fails and you're going to have a problem.Like simply taking out the brain the body becomes useless.Or taking out part of the body and the rest is at a disadvantage.If you have a 100 mini drones go into battle using a network that has a plan for every single 1 of those drones and then say 40% of them fail or the "network" fails you're going to have a problem.The units which cannot function properly and would not be able to improvise.That means AI is the thing that will take drone technology to where it can replace humans in that sense.But the problem is AI is currently to dumb,to slow or too big to be used.Not yet anyway it's going to take time.I hardly think you want me to give you a lesson in network or software programming so i made it short.

SingaporeTyrannosaur SG

Try to imagine it in an american drawl, and you can see that it is hardly impossible.

You seem to doubt the source simply because it contradicts you. After all, the pilot is stating something obvious to all who have studied air combat, that it is more about violent manuvering, and hasty moves and counter moves to get the direction right instead of the clear-headed tatical game of wits you suggested. Air combat is a snappy and brutally simple thing, and unless you can play chess with 0.3 second turns, it is going to be hardly anything we would define as humanisticly refined. As a matter of fact, close in air combat would is inherently a messy, animalistic affair, probably why they call it "dogfighting" and "furball" rather then "airborne chess".

SingaporeTyrannosaur SG

Erm, I think you don't understand the new model a bit at all, based on what you have said. The entire premise of this information sharing network is it's invulnerability and reduncancy to damage. When you kill a single ant, or even a large number of ants in the colony, the colony's overall C&C is hardly affected at all (as a matter of fact, the colony can actually be run and recover on one worker if need be). While if you de-liver a human, he's dead meat. That's the diff between the old and the new model.

The only way to distrupt this network is two options:either you kill every drone in the network, or kill the guy giving the overall orders (in this case you have to nuke the Pentagon or something like that). Either ways, these are vulnerabilites that exist for all weapons systems, manned or not.

« Previous Page Next Page »