Since we had this discussion about UAVs and F22s etc...

Discuss life, the universe, and everything with other members of this site. Get to know your fellow polywell enthusiasts.

Moderators: tonybarry, MSimon

palladin9479
Posts: 388
Joined: Mon Jan 31, 2011 5:22 am

Re: Since we had this discussion about UAVs and F22s etc...

Post by palladin9479 »

ltgbrown wrote:I wonder when the last aircraft carrier will be bought? Imagine a strike group with an E-2D system in an aerostat overhead (24/7 by the way) and a fleet of drones (that could have substantially different takeoff and landing requirements, thus smaller and denser flight decks). Would we still need Nimitz/Ford class size aircraft carriers?

CDR Glenn "Woodgie" Brown
US Naval Aviator
3500 flight hours
750 arrested landings
Former Mini Boss
They will just modify future carrier design's, and we'll always have manned aircraft just not as many. Things like CAS and interdiction / air superiority are things that require manned pilots while surgical strikes and surveillance lend themselves very well to unmanned remotely operated vehicles. The #1 biggest problem with remotely piloting something is network latency, satellite systems have latency times measured in the hundreds of milliseconds while radio requires line of sight or multiple relays, while a pilot in the cockpit has zero network latency. "Split second" real time decisions simply aren't possible with such high latency remote connections. The only way around that would be to utilize artificial intelligence and give the drone the capacity to initiate it's own target and kill order, think long and hard about that one.

I can completely see a small fleet of localized drones that stay within a certain range of a carrier and act as a shield against potential threats while a few are sent out on long range strikes or surveillance. That would leave human pilots to be long range interceptors or to provide CAS during a ground firefight.

Skipjack
Posts: 6805
Joined: Sun Sep 28, 2008 2:29 pm

Re: Since we had this discussion about UAVs and F22s etc...

Post by Skipjack »

palladin9479 wrote:
ltgbrown wrote:I wonder when the last aircraft carrier will be bought? Imagine a strike group with an E-2D system in an aerostat overhead (24/7 by the way) and a fleet of drones (that could have substantially different takeoff and landing requirements, thus smaller and denser flight decks). Would we still need Nimitz/Ford class size aircraft carriers?

CDR Glenn "Woodgie" Brown
US Naval Aviator
3500 flight hours
750 arrested landings
Former Mini Boss
They will just modify future carrier design's, and we'll always have manned aircraft just not as many. Things like CAS and interdiction / air superiority are things that require manned pilots while surgical strikes and surveillance lend themselves very well to unmanned remotely operated vehicles. The #1 biggest problem with remotely piloting something is network latency, satellite systems have latency times measured in the hundreds of milliseconds while radio requires line of sight or multiple relays, while a pilot in the cockpit has zero network latency. "Split second" real time decisions simply aren't possible with such high latency remote connections. The only way around that would be to utilize artificial intelligence and give the drone the capacity to initiate it's own target and kill order, think long and hard about that one.

I can completely see a small fleet of localized drones that stay within a certain range of a carrier and act as a shield against potential threats while a few are sent out on long range strikes or surveillance. That would leave human pilots to be long range interceptors or to provide CAS during a ground firefight.
I always imagined manned aircraft and drones to be operating in a group. There would be a few manned aircraft provide support and some guidance and a large group of drones does the brunt of the work. Then you simply overwhelm the enemy with numbers. One on one drones might be inferior (at least in the near term) but if the ratio is 5 to one or more, the drones will most likely win.
Also want to point out that LEO constellations, such as the ones planned by Musk would have a much lower latency, more comparable to land lines.
And finally it is worth noticing that with self guided missiles that autonomously update targeting information, we are already pretty close to that blurry(!) line of where autonomous killing begins, if we have not crossed it already.
https://www.youtube.com/watch?v=SEohrQDC7vo
Last edited by Skipjack on Sat Apr 25, 2015 7:25 pm, edited 2 times in total.

Skipjack
Posts: 6805
Joined: Sun Sep 28, 2008 2:29 pm

Re: Since we had this discussion about UAVs and F22s etc...

Post by Skipjack »

This is relevant in the context:
https://www.youtube.com/watch?v=tjlPoD_9BJs

choff
Posts: 2447
Joined: Thu Nov 08, 2007 5:02 am
Location: Vancouver, Canada

Re: Since we had this discussion about UAVs and F22s etc...

Post by choff »

CHoff

palladin9479
Posts: 388
Joined: Mon Jan 31, 2011 5:22 am

Re: Since we had this discussion about UAVs and F22s etc...

Post by palladin9479 »

Skipjack wrote:
palladin9479 wrote:
ltgbrown wrote:I wonder when the last aircraft carrier will be bought? Imagine a strike group with an E-2D system in an aerostat overhead (24/7 by the way) and a fleet of drones (that could have substantially different takeoff and landing requirements, thus smaller and denser flight decks). Would we still need Nimitz/Ford class size aircraft carriers?

CDR Glenn "Woodgie" Brown
US Naval Aviator
3500 flight hours
750 arrested landings
Former Mini Boss
They will just modify future carrier design's, and we'll always have manned aircraft just not as many. Things like CAS and interdiction / air superiority are things that require manned pilots while surgical strikes and surveillance lend themselves very well to unmanned remotely operated vehicles. The #1 biggest problem with remotely piloting something is network latency, satellite systems have latency times measured in the hundreds of milliseconds while radio requires line of sight or multiple relays, while a pilot in the cockpit has zero network latency. "Split second" real time decisions simply aren't possible with such high latency remote connections. The only way around that would be to utilize artificial intelligence and give the drone the capacity to initiate it's own target and kill order, think long and hard about that one.

I can completely see a small fleet of localized drones that stay within a certain range of a carrier and act as a shield against potential threats while a few are sent out on long range strikes or surveillance. That would leave human pilots to be long range interceptors or to provide CAS during a ground firefight.
I always imagined manned aircraft and drones to be operating in a group. There would be a few manned aircraft provide support and some guidance and a large group of drones does the brunt of the work. Then you simply overwhelm the enemy with numbers. One on one drones might be inferior (at least in the near term) but if the ratio is 5 to one or more, the drones will most likely win.
Also want to point out that LEO constellations, such as the ones planned by Musk would have a much lower latency, more comparable to land lines.
And finally it is worth noticing that with self guided missiles that autonomously update targeting information, we are already pretty close to that blurry(!) line of where autonomous killing begins, if we have not crossed it already.
https://www.youtube.com/watch?v=SEohrQDC7vo
LEO orbit is still too long a latency and no where close to landlines unless you inject a ton of routers / infrastructure into the land line while giving the LEO route almost none. It's the raw distance involved that screws up realized latency. A ground station to bird back down to ground (or near enough) is approximately 750~1000 ms depending on if it has to bounce to other birds or not first. Land line will get the same at 70~250ms depending on distance and number of nodes it must travel first.

Smart missiles aren't fire and forget, they need to have a target programmed into them first and thus target selection and kill order are given by a human before hand. I wouldn't see a problem programming a done to autonomously fly to location X and destroy target Y then fly back, I see a problem with the drone flying autonomously to location X and then deciding what it should kill then flying back. That is why drones are perfect for surveillance and strategic strikes, those are things where we can program in the necessary information ahead of time while CAS and air interdiction require target determination, acquisition and execution to be done in real time upon arriving at the given location. And yeah I envision a future where you have a set of human pilots controlling one aircraft that simultaneously controls a dozen or more local drones. The backseat bomber position just transforms into a drone specialist that utilizing the aircraft as a C2 hub and directs the drones by giving them specific programs (defensive / offensive) to execute. Would lead to some interesting combat scenarios where the victor is the one who can most successfully employ their drone arsenal in the local airspace. The front seat pilot's job is to keep the back seat guy alive and provide them with the best angle or position to control the battlespace.

hanelyp
Posts: 2261
Joined: Fri Oct 26, 2007 8:50 pm

Re: Since we had this discussion about UAVs and F22s etc...

Post by hanelyp »

A manned aircraft commanding a flight of unmanned aircraft which in turn protect the manned aircraft can put a human in the loop to select targets with minimal latency, subject to the limited bandwidth of the human. Once a target is selected the automated aircraft can use superior numbers to overwhelm defenses.
The daylight is uncomfortably bright for eyes so long in the dark.

palladin9479
Posts: 388
Joined: Mon Jan 31, 2011 5:22 am

Re: Since we had this discussion about UAVs and F22s etc...

Post by palladin9479 »

Yep and you can have dozens of operational programs stored that the human operator selects from and directs to the drone. Stuff like intercepting incoming threats, solo or group attacking individual targets, attacking all local targets that fit a specific profile, and so forth. Then you introduce electronics warfare where each system attempts to hack or jam the other, gets into some pretty futuristic stuff.

GIThruster
Posts: 4686
Joined: Tue May 25, 2010 8:17 pm

Re: Since we had this discussion about UAVs and F22s etc...

Post by GIThruster »

palladin9479 wrote:LEO orbit is still too long a latency. . . is approximately 750~1000 ms.
Can you please explain this? Certainly the time to send light 400 miles round trip is nothing like a second.
Smart missiles aren't fire and forget, they need to have a target programmed into them first and thus target selection and kill order are given by a human before hand.
This is simply not true. Smart missiles launch before they have a target and acquire their targets in flight. The much greater range radar aboard the aircraft finds and distinguishes viable targets, and then launches, but the missiles are blind until they get within much closer range and their onboard radar reacquires. It is the severely limited range of the radar aboard each missile that requires they be "smart" enough to find and acquire targets once they're in range. The Pentagon is considering extending this method of target acquisition by moving to much greater range missiles carried aboard missile gunboats like the B1. That platform could carry missiles on deep penetration missions with strike craft such as X-47b's and have hundreds of miles range on anything that lofts into the air to meet the raiders.

F-22's are necessary right now, but they will not be for much longer. Smarter, longer range missiles flying cover for drones is a pretty good scenario for success. And dogfighting is already dead. Missiles killed it more than a decade ago. There's no reason to pretend we'll always have men aboard fighters. It just ain't so. Eventually they'll take them out of the missile carriers too.
"Courage is not just a virtue, but the form of every virtue at the testing point." C. S. Lewis

Skipjack
Posts: 6805
Joined: Sun Sep 28, 2008 2:29 pm

Re: Since we had this discussion about UAVs and F22s etc...

Post by Skipjack »

SpaceX LEO constellation would be at 1200 km altitude. Speed of light is ~300,000 km a second. With an angle or another satellite for relay, we can maybe get to 10,000 km total distance. That means a latency of 33 ms, nowhere near 1 second. I don't think even the GTO constellations have one second lag (more like 500 ms).
I have internet with Comcast here and when I do a trace route to comcast.com, I get at least 200ms latency. Doing a Speedtest.net to Comcast's nearest speed test server just 10 miles away gives me 11ms latency...

GIThruster
Posts: 4686
Joined: Tue May 25, 2010 8:17 pm

Re: Since we had this discussion about UAVs and F22s etc...

Post by GIThruster »

Even if you are bouncing to Geo, that's 35k km each way, or 70k km. . .no where near a second. When you look at the vast difference between AI and human reaction times, drones are probably much faster than human pilots, especially given they can pull 12 gee maneuvers and humans generally about 4 max.
"Courage is not just a virtue, but the form of every virtue at the testing point." C. S. Lewis

Skipjack
Posts: 6805
Joined: Sun Sep 28, 2008 2:29 pm

Re: Since we had this discussion about UAVs and F22s etc...

Post by Skipjack »

Meteor BVRAAM with 100 km range...
https://www.youtube.com/watch?v=DiinNwfNi1w

palladin9479
Posts: 388
Joined: Mon Jan 31, 2011 5:22 am

Re: Since we had this discussion about UAVs and F22s etc...

Post by palladin9479 »

Skipjack wrote:SpaceX LEO constellation would be at 1200 km altitude. Speed of light is ~300,000 km a second. With an angle or another satellite for relay, we can maybe get to 10,000 km total distance. That means a latency of 33 ms, nowhere near 1 second. I don't think even the GTO constellations have one second lag (more like 500 ms).
I have internet with Comcast here and when I do a trace route to comcast.com, I get at least 200ms latency. Doing a Speedtest.net to Comcast's nearest speed test server just 10 miles away gives me 11ms latency...
It's ... a lot more complicated then that.

Remember signal propagation along a medium is almost the speed of light, yet we don't get anywhere near that kind of speed along the total signal path because your sending far more then a single waveform. It's a bunch of waves bunched together and compressed in a packet that represents a datagram, you need tens of thousands or more of these datagrams per second per connection. At each stop there is a PHY adapter that takes the physical (layer 1) signal and converts it into a different form for temporary analysis and storage / buffering while it's waits to be transmitted onto it's next path. Prior to transmission it needs to be converted from it's temporal format to another physical one based on whatever form it's being transmitted as (light / radio signal / copper signal) and sent on it's merry way. The actual time spent in between transmission points is minimal compared to how long it's spent inside the buffers at the ends of that transmission point.

The reason for ridiculously high latency on satellite connections isn't so much the raw distance, though that doesn't help, but the sheer amount of buffering and converting that needs to happen on both sides as it's rarely as straight a path as an Ethernet connection. The reason you see so much latency on your shitty comcast connection is because they have complete crap infrastructure with overloaded core infrastructure. From my apartment in South Korea I have an 8ms ping between me and Google's servers in Seoul which is approximately 300km away. On the other hand I've worked with DAMA based K-band satellite systems that had a 750~900ms ping time from client to bird(s) to base station and a nearly 2000ms ping doing an M hop from client to another client at a different site.

Anyhow your never going to see real time control of a drone unless your very close to it, what you get instead is called near-real time which is typically ~1s or so behind. That's good enough to guide it in, and do counter-insurgency stuff like airstrikes and surveillance but not for Close Air Support and air interdiction where you need to receive information, make a decision and execute that decision within a quarter to half second time frame.

krenshala
Posts: 914
Joined: Wed Jul 16, 2008 4:20 pm
Location: Austin, TX, NorAm, Sol III

Re: Since we had this discussion about UAVs and F22s etc...

Post by krenshala »

I agree with what palladin9479 is stating on network connectivity, as it matches what I've learned first hand over the last 25 years doing network support/administration.

For a better idea of how long range drone connections will work, I would like to point out any online game where you have a latency of 150ms or longer. While the time difference between when you give a command and when it executes can be compensated for by a skilled user (or player, in the case of games) it can be extremely difficult to be accurate and timely when it takes twice that time to see what the results of the command are and react accordingly. Not only that, but if your network latency is near human reaction time (~200ms) then you are basically going to have remotely controlled drones that take twice as long to react to a new situation. If being shot at, that time (~400ms) could be more than enough to lose the drone before you can do anything about it.

If you are close enough to reduce the network latency to 'near real-time' levels, you are almost guaranteed to be close enough to worry about getting into the thick of things personally, and not just remotely through a drone.

paperburn1
Posts: 2484
Joined: Fri Jun 19, 2009 5:53 am
Location: Third rock from the sun.

Re: Since we had this discussion about UAVs and F22s etc...

Post by paperburn1 »

sooo that when the swarm defense becomes a player
I am not a nuclear physicist, but play one on the internet.

krenshala
Posts: 914
Joined: Wed Jul 16, 2008 4:20 pm
Location: Austin, TX, NorAm, Sol III

Re: Since we had this discussion about UAVs and F22s etc...

Post by krenshala »

I see it as picking your problem: AI that may not be able to cope well enough to give the human time to assess and respond versus having a human close enough to respond in a timely fashion and thus is also close enough to have to worry about the outcome.

I'm sure there is a middle ground that will make drones viable in a number of roles, the question is what roles, and what limitations will those roles require to be functionally useful.

Post Reply