Technology Stats Help

[KaiserMelech Mikhail]

[quote name='HHAYD' timestamp='1286070695' post='2472662']
Is it because of limitation in computing power?
[/quote]
Mixed with an absence of any realistic AI.

[Sargun II]

[quote]The MoD was keen to stress that any operational vehicle would not leave human control.

‘It is designed to be, at all times, under the control of a highly trained military operator on the ground and can be operated remotely,’ said Gerald Howarth, minister for international security strategy (ISS).
[/quote]

Taranis requires human operation at all time, and all it proves is that there is a concept. MANTIS doesn't even say what the "missions" are, so that link is useless. Something that flies without human intervention is one thing, being a fully operational autonomous aircraft is another. The popularmechanics article puts 2025 as a date on it, which is reaching (and still five years above the cap). The X-45A was [i]always[/i] managed by a pilot-operator, and the Phantom Ray is only a demonstrator aircraft.

So what we have here is that there is a concept, but nothing is proven and that human intervention is required for advanced tasks at all times. Further, the best "date" that can be reached is 2025, and historically we've been rather slow on meeting deadlines - so not only is the base date far past the cap, but the actual date will most likely be far past it as well. The answer is still [b]no[/b], you cannot have them.

[iKrolm]

[quote name='HHAYD' timestamp='1286070695' post='2472662']
Is it because of limitation in computing power?
[/quote]

No, it's because the technology won't be developed that fast: it costs a [i]lot[/i] of money to build & get approved anything that flies and there isn't the need to spend that money right now while remotely operated planes seem to work just fine.

[HHAYD]

Could I at least RP automated aircraft that only do recon and enemy aircraft detection operations?

Edited October 3, 2010 by HHAYD

[Sargun II]

Make it so that there is always somebody that can take over if there is a problem and I'll be fine with it.

[HHAYD]

[quote name='Sargun' timestamp='1286076236' post='2472714']
Make it so that there is always somebody that can take over if there is a problem and I'll be fine with it.
[/quote]
If a computer learns how to do new things by experimenting and watching, can adapt to new situations without massive amount of programming, and upload its files to other computers so they can also learn without having to repeat the same thing, would it be realistic for the automated aircraft to slowly reduce their reliance on humans?

http://www.wpafb.af.mil/news/story.asp?id=123095781

http://www.walyou.com/blog/2008/06/12/self-learning-robot-solves-mazes/

Self-learning robots, give it basic coding and commands and then let it write the rest of it itself based on what it experiences.

http://www.popsci.com/technology/article/2010-09/video-using-new-learning-algorithm-archer-bot-learns-how-aim-and-shoot-bow-and-arrow

A robot learns how to shoot arrows instead of directly programming it to do it.

http://www.popsci.com/gadgets/article/2010-09/using-new-algorithm-emotional-computers-can-respond-aesops-fables

Computers that can experience emotions and free-thinking.

http://www.popsci.com/scitech/article/2009-08/evolving-robots-learn-lie-hide-resources-each-other
http://www.popsci.com/technology/article/2010-09/deceptive-robots-learn-when-and-how-pull-fast-one

And at a certain point with the correct codes, it can learn to lie after evolving via software. Though that isn't something I would want in an automated aircraft, unless if it is tricking enemy pilots.

http://www.science20.com/news_releases/automated_helicopter_teaches_itself_to_fly

Automated helicopters that can teach themselves how to fly and pull off complex maneuvers by simply watching other helicopters operate. All it required was programing on how to learn and operate itself.

http://www.popsci.com/technology/article/2010-08/willow-garages-pr2-hits-another-laundry-landmark-time-pairing-your-socks

A robot that learned how to play pool, fetch (and open) beer bottles, and properly pairing socks.

http://www.popsci.com/technology/article/2010-08/darpa-funded-endeavor-creates-probability-processing-chip

A processor chip that runs on probability rather than hard binary (1s and 0s). Useful for predicting what is the enemy pilot is planning on doing with less programming.

Edited October 3, 2010 by HHAYD

[iKrolm]

[quote name='HHAYD' timestamp='1286076962' post='2472719']
If a computer learns how to do new things by experimenting and watching, can adapt to new situations without massive amount of programming, and upload its files to other computers so they can also learn without having to repeat the same thing, would it be realistic for the automated aircraft to slowly reduce their reliance on humans?

http://www.wpafb.af.mil/news/story.asp?id=123095781

http://www.walyou.com/blog/2008/06/12/self-learning-robot-solves-mazes/

Self-learning robots, give it basic coding and commands and then let it write the rest of it itself based on what it experiences.

http://www.popsci.com/technology/article/2010-09/video-using-new-learning-algorithm-archer-bot-learns-how-aim-and-shoot-bow-and-arrow

A robot learns how to shoot arrows instead of directly programming it to do it.

http://www.popsci.com/gadgets/article/2010-09/using-new-algorithm-emotional-computers-can-respond-aesops-fables

Computers that can experience emotions and free-thinking.

http://www.popsci.com/scitech/article/2009-08/evolving-robots-learn-lie-hide-resources-each-other
http://www.popsci.com/technology/article/2010-09/deceptive-robots-learn-when-and-how-pull-fast-one

And at a certain point with the correct codes, it can learn to lie after evolving via software. Though that isn't something I would want in an automated aircraft, unless if it is tricking enemy pilots.

http://www.science20.com/news_releases/automated_helicopter_teaches_itself_to_fly

Automated helicopters that can teach themselves how to fly and pull off complex maneuvers by simply watching other helicopters operate. All it required was programing on how to learn and operate itself.

http://www.popsci.com/technology/article/2010-08/willow-garages-pr2-hits-another-laundry-landmark-time-pairing-your-socks

A robot that learned how to play pool, fetch (and open) beer bottles, and properly pairing socks.

http://www.popsci.com/technology/article/2010-08/darpa-funded-endeavor-creates-probability-processing-chip

A processor chip that runs on probability rather than hard binary (1s and 0s). Useful for predicting what is the enemy pilot is planning on doing with less programming.
[/quote]


For the love of god, Popular Science is not a reliable source.

Edited October 3, 2010 by iKrolm

[Sargun II]

Robots programmed to do specific simple things =/= robots that can evolve and think for themselves. The robots were supposed to do X, and they did X after several failures or extensive algorithms that was from programming designed to match trajectories, something that has been out for many years. Some of those articles and headlines are completely misleading as well. The things you provided are completely out of context and does not change my mind.

Edited October 3, 2010 by Sargun

[king of cochin]

[quote name='HHAYD' timestamp='1286047423' post='2472391']
Would adding thrust vectoring to the jet engines would be feasible?
[/quote]

Thrust vectoring is a technology used to increase maneuverability, yes. But they too produce immense lateral stress on the air frame when moving at very high speeds. In fact the fastest of airplanes usually retract their wings to reduce drag once they go beyond the Mach speeds. Beyond a certain speed lift is no longer the issue, drag is. Bring on the issue of lateral strain due to maneuvers and the structural stability is totally gone.

[HHAYD]

[quote name='king of cochin' timestamp='1286083183' post='2472795']
Thrust vectoring is a technology used to increase maneuverability, yes. But they too produce immense lateral stress on the air frame when moving at very high speeds. In fact the fastest of airplanes usually retract their wings to reduce drag once they go beyond the Mach speeds. Beyond a certain speed lift is no longer the issue, drag is. Bring on the issue of lateral strain due to maneuvers and the structural stability is totally gone.
[/quote]
I understand, I won't RP my aircraft attempting to fly at such high speed while trying to turn. But at 1-3 mach, I suppose they would be highly maneuverable since they are small compared to all other modern manned military jets.
[quote name='iKrolm' timestamp='1286078711' post='2472747']
For the love of god, Popular Science is not a reliable source.
[/quote]
What about the videos of them?

Robot learns how to walk by itself and adapt to new terrains realistically: http://news.bbc.co.uk/2/hi/6291746.stm
[quote]Professor Woergoetter, of the University of Gottingen, in Germany, said: "When RunBot first encounters a slope these low level control circuits 'believe' they can continue to walk up the slope without having to change anything.

"But this is misguided and as a consequence the machine falls backwards. This triggers the other sensors and the highest loop we have built into RunBot - the learning circuitry - and from that experience of falling the machine knows that something needs to be changed." [/quote]
http://www.youtube.com/watch?v=lYANWOd6DyE&feature=related
http://www.youtube.com/watch?v=NPdNDjV_9TE&feature=related

http://dsc.discovery.com/news/2006/11/16/starfish_tec.html?category=technology

[quote]The software in conventional robots typically doesn't account for unpredictable changes in the environment, or to the robot itself, that could restrict its movement.

But the Starfish can explore its own abilities and limitations, taking them into account before planning a move.

The robot begins by getting a sense of itself, testing each of its joints with random motions. Sensors on the joints capture each joint's range of motion and feed that information to the 15 mathematical models built into its controlling software.

Each model figures out one possible mode of locomotion. For example, one model might find how the robot is capable of scurrying scorpion-like on three legs, using the fourth like a tail for balance. Another might offer a way for the robot to scuttle sideways like a crab.[/quote]
http://blogs.discovermagazine.com/sciencenotfiction/2010/08/25/the-new-ai-turn-robots-into-infant-scientists/
[quote]It worked something like this: first, make an arbitrary movement. What this means is that the robot sends out signals to its body, without knowing what those signals will do. While sending these movement signals out, the robot records sensory signals that tell it about what happened to the body due to that movement (scientific process analog: experiment)[/quote]

[quote]The starfish robot shown in the photo has had one of its arms pulled off, and after a brief learning process, it figures out its new body shape and saunters off![/quote]
http://www.technologyreview.com/blog/editors/25799/?ref=rss
Same archery robot, but with a video of it learning.

http://www.itnews.com.au/News/232971,researcher-builds-machines-that-daydream.aspx
Same emotional feeling robot...
[quote]The machine freely associated three stories: The Thirsty Pigeon; The Cat and the &#$@; and The Wolf and the Crane.

When queried on the association, the machine responded: "I felt sad for the bird."[/quote]

http://discovermagazine.com/2008/jan/robots-evolve-and-learn-how-to-lie
[quote]By the 50th generation, the robots had learned to communicate—lighting up, in three out of four colonies, to alert the others when they’d found food or poison. The fourth colony sometimes evolved “cheater” robots instead, which would light up to tell the others that the poison was food, while they themselves rolled over to the food source and chowed down without emitting so much as a blink.

Some robots, though, were veritable heroes. They signaled danger and died to save other robots.[/quote]

http://www.eurekalert.org/pub_releases/2010-09/giot-rgr090910.php

"""The hider's set of false communications was defined by selecting a pattern of knocked over markers that indicated a false hiding position in an attempt to say, for example, that it was going to the right and then actually go to the left," explained Wagner.

The hider robots were able to deceive the seeker robots in 75 percent of the trials, with the failed experiments resulting from the hiding robot's inability to knock over the correct markers to produce the desired deceptive communication.""

http://www.science20.com/news_releases/automated_helicopter_teaches_itself_to_fly
[quote]The dazzling airshow is an important demonstration of "apprenticeship learning," in which robots learn by observing an expert, rather than by having software engineers peck away at their keyboards in an attempt to write instructions from scratch.

Writing software for robotic helicopters is a daunting task, in part because the craft itself, unlike an airplane, is inherently unstable. "The helicopter doesn't want to fly. It always wants to just tip over and crash" said Oku, the pilot.[/quote]

http://www.pcworld.com/article/203541/chip_startup_developing_probability_processor.html
[quote]"These circuits can accept inputs and calculate outputs that are between 0 and 1, directly representing probabilities -- levels of certainty," Lyric said, adding that its probability processors are designed to perform calculations in parallel, rather than in sequential order like a traditional CPU.[/quote]

The automated aircraft that I would RP would attempt to test out every system to see what each signal does and then attempt to fly after watching several videos of aircraft flying. Then it would be fed videos of fighter jets pulling highly complex maneuvers and the automated aircraft will attempt to reproduce the results. Part(s) of it will be taken off and the aircraft will attempt to find a workaround to fly without such part, such as most of a wing. After that, it will be put into a practice aerial fighting, pitting it against standard UAVs and then manned fighter jets until ready for actual combat. There will be numerous crashes during the aircraft' experimentation though, especially in its early stages.

I plan on RPing the aircraft taking about six RL months to be combat ready. Even during combat, it will behave like a beginner pilot and fare poorly, and after about a total of six RL months of combat and pushing their computing power to the limits, it will match the skills of decent pilots. That's a total of 12 RP years of learning, much longer than the adaptive robots took to accomplish their tasks or pilots took to learn how to be a decent jet fighter pilot.

Edited October 3, 2010 by HHAYD

[Sargun II]

How will it identify friends and foes? Robots do not see as we do - all they see comprises of geometric shapes and colors, and even the smartest robots have trouble doing simple things such as identifying what water is. Programming a robot to learn how to fight in the air without any human intervention would require extremely flawless code and possibly decades of R&D to get it off the ground. We still have 15 years for semi-autonomous planes (at BEST), and I'm going to go out on a limb and say it'll take more than twice that to get fully automated planes fit for combat, if at all.

The answer is still no.

[king of cochin]

[quote name='HHAYD' timestamp='1286113261' post='2472948']
I understand, I won't RP my aircraft attempting to fly at such high speed while trying to turn. But at 1-3 mach, I suppose they would be highly maneuverable since they are small compared to all other modern manned military jets.

[/quote]

Then of course thrust vectoring in an unmanned combat aircraft would be excellent. They would indeed be highly maneuverable and there wont be the usual limit on G force that hampers a manned fighter. The lack of an aviator also means all those life support systems are not necessary, thus greatly lightening the aircraft and improving its capabilities.

[HHAYD]

[quote name='Sargun' timestamp='1286122171' post='2473022']
How will it identify friends and foes? Robots do not see as we do - all they see comprises of geometric shapes and colors, and even the smartest robots have trouble doing simple things such as identifying what water is. Programming a robot to learn how to fight in the air without any human intervention would require extremely flawless code and possibly decades of R&D to get it off the ground. We still have 15 years for semi-autonomous planes (at BEST), and I'm going to go out on a limb and say it'll take more than twice that to get fully automated planes fit for combat, if at all.

The answer is still no.
[/quote]
How do manned aircraft tell friend from foe if the detected aircraft is outside of visual range?

[Sargun II]

[quote name='HHAYD' timestamp='1286128241' post='2473125']
How do manned aircraft tell friend from foe if the detected aircraft is outside of visual range?
[/quote]

So you didn't answer my question, thanks.

The answer is no, the answer will always be no.

[HHAYD]

[quote name='Sargun' timestamp='1286128671' post='2473136']
So you didn't answer my question, thanks.

The answer is no, the answer will always be no.
[/quote]
So manned aircraft can't detect other manned aircraft outside of the visual range? Here's a hint, its found in almost all operational modern jets and there's a reason why we don't have missiles locking onto the aircraft that they were fired from.

Edited October 3, 2010 by HHAYD

[king of cochin]

[quote name='Sargun' timestamp='1286122171' post='2473022']
How will it identify friends and foes? Robots do not see as we do - all they see comprises of geometric shapes and colors, and even the smartest robots have trouble doing simple things such as identifying what water is. Programming a robot to learn how to fight in the air without any human intervention would require extremely flawless code and possibly decades of R&D to get it off the ground. We still have 15 years for semi-autonomous planes (at BEST), and I'm going to go out on a limb and say it'll take more than twice that to get fully automated planes fit for combat, if at all.

The answer is still no.
[/quote]

The unmanned aircraft could still be programmed to engage targets based on IFF interrogation, however point remains that Artificial Intelligence needed to operate a UCAV of this level is still far ahead in the future. Robotics is only now seeing AI with 3 dimensional perception, for aviation there are far more parameters that need to be observed at any moment in time and a suitable architecture for that level of AI is not possible within the next 15-20 years.

The best bet would be to use them as remotely piloted aircrafts.

[HHAYD]

[quote name='king of cochin' timestamp='1286159395' post='2473758']
The unmanned aircraft could still be programmed to engage targets based on IFF interrogation, however point remains that Artificial Intelligence needed to operate a UCAV of this level is still far ahead in the future. Robotics is only now seeing AI with 3 dimensional perception, for aviation there are far more parameters that need to be observed at any moment in time and a suitable architecture for that level of AI is not possible within the next 15-20 years.

The best bet would be to use them as remotely piloted aircrafts.
[/quote]
If the automated aircraft are going up against enemy aircraft that were recorded in the database before, then they can easily identify and target them by looking at the shapes (calculated with multiple AESA, non-visible laser transmitters/receivers, and other equipments from other automated aircraft) and IFF. But for friendly aircraft that are the exact same copy of the hostile ones, then the automated aircraft can only rely on IFF.

For aircraft that do not match either friendly or enemy IFF or are not in the database, the manned fighter aircraft will take care of that issue.

To cut down on the programming, the automated aircraft could be put into practices where they have to distinguish between friendly and enemy aircraft and programmed to go for the highest points, enemy aircraft shot down, +1, friendly aircraft shot down, -10, shooting down civilian aircraft, -10. Though it will delay the deployment of the automated aircraft into the sky during war.

[Subtleknifewielder]

The point is, HHAYD, the smartest AI we have today, I believe, is as smart as...a rat. Now while a rat is pretty cunning for a rodent...it's still a rodent.

[HHAYD]

[quote name='Subtleknifewielder' timestamp='1286213980' post='2474305']
The point is, HHAYD, the smartest AI we have today, I believe, is as smart as...a rat. Now while a rat is pretty cunning for a rodent...it's still a rodent.
[/quote]
Our smartest AIs so far were directly coded, a very hard and error-prone task. Nor are they capable of sending information to each other unlike the AI system that I planned on using. My automated aircraft are mostly harmless when flying alone, but in squadrons that are all interconnected via wireless, it would be suicidal to attack them even with the same number of the manned aircraft.

A single neuron cell is weak, but many of them interconnected...

For example, human infants are not very bright. But give them time to grow and learn, and usually they will become smarter. Same with the adaptive AI system and I plan on RPing it for about 2 RL years (24 RP years, depending on how much combat time they had) from the development to the point where it can fight toe-to-toe battles against average fighter pilots. Then they will be transferred to more advanced automated aircraft to allow them to continue learning when their computing power are pushed to the limit.

However, I agree they will never be able to stand toe-to-toe battles against expert fighter pilots without needing reinforcements. But it would be very hard to man an air force consisting of entirely expert human pilots. A RPer (such as Lavo or Lyth) with superior tech and a larger air force or a RPer who spent large amount of time making numerous detailed posts about the training of his/her pilots could still kick even experienced automated aircraft's rear ends.

To ensure other RPers do not abuse automated aircraft, I decided to count every 4 of my IG aircraft as 3 RP automated aircraft squadrons instead of 4. This will force them to decide if they want to use automated aircraft (which would reduce their max air force size) or stick with manned and remote controlled aircraft. Is that fair enough?

Edited October 5, 2010 by HHAYD

[JerreyRough]

I'd doubt that the point when AI is doing the fighting for us will be coming anytime soon; but regarding the amount of time you're going to consitantly RP this progress, I would let it fly personally. Although keep in mind that this will mostly just make RP more interesting; not actually give you a big advantage. At least, I think so anyways. Yet still an advantage anyways, as it would take time and effort to do.

Keep in mind that such "learning" programs are dangerous things, and I'm not talking about evil movie AI's. They learn by making, writing and rewriting their own code; you could say that humans do the same thing, but we've had a long time to work out the kinks. The dangerous part of this is that they possibly can redefine things that you think were hard-wired to not be re-definable. They can create new errors, bugs, mis-fire their weapons or won't do anything when ordered to. They can easily break themselves in this learning process and possibly render years of work useless. Point is, if it weren't for this issues then we'd have far more advanced AI technology.

But I didn't write that to stop you; rather, it's to give you reasons for why it takes so long and how you'll get around those issues.

[Sargun II]

The answer is still [b]no.[/b]

If the cap gets changed to 2050 you'll get your fully-automated-but-not-as-good-as-piloted planes.

[HHAYD]

[quote name='JerreyRough' timestamp='1286249157' post='2475363']
I'd doubt that the point when AI is doing the fighting for us will be coming anytime soon; but regarding the amount of time you're going to consitantly RP this progress, I would let it fly personally. Although keep in mind that this will mostly just make RP more interesting; not actually give you a big advantage. At least, I think so anyways. Yet still an advantage anyways, as it would take time and effort to do.

Keep in mind that such "learning" programs are dangerous things, and I'm not talking about evil movie AI's. They learn by making, writing and rewriting their own code; you could say that humans do the same thing, but we've had a long time to work out the kinks. The dangerous part of this is that they possibly can redefine things that you think were hard-wired to not be re-definable. They can create new errors, bugs, mis-fire their weapons or won't do anything when ordered to. They can easily break themselves in this learning process and possibly render years of work useless. Point is, if it weren't for this issues then we'd have far more advanced AI technology.

But I didn't write that to stop you; rather, it's to give you reasons for why it takes so long and how you'll get around those issues.
[/quote]
I know it won't give me an advantage over manned aircraft, I wanted a type of air force that suit my tastes, a highly maneuverable, stealthy (at low or mid speed), can communicate with other aircraft easily, and yet fast aircraft. Range, firepower, and numbers are secondary.

I am considering two ways of allowing the AIs to learn, one in an actual aircraft. I would have more than one automated aircraft (with ruggized computers in armored shells), and they will learn the mistakes that lead to the mechanical/software failure of another aircraft.

Another way would be in a virtual setting so no money is lost on repairs or replacements, but that would spawn additional errors.

Edited October 5, 2010 by HHAYD

[Kankou]

Suggestion: Radio control.

[Captain Enema]

Weeeeeeeeeeeeee I'm a pilot now mommy!


suggestion: Get 12 year old boys to fly your UAVs.

Epic hilarity to follow.

[HHAYD]

[quote name='Kankou' timestamp='1286264651' post='2475727']
Suggestion: Radio control.
[/quote]
There's always a delay between signal sent and received and it gets worse when the aircraft are being connected via satellite, on the edge of the radio connectivity range, or there is a bad weather interfering with the signal, resulting in aircraft that act as if they were being piloted by drunken pilots, always several seconds too late to react.

Plus, radio signals can be jammed, resulting in dead brain aircraft.

Edited October 5, 2010 by HHAYD