Unmanned Aircraft Crosses the Atlantic

[Foxbat]

I've been thinking of the future of unmanned aircraft and I think there may be a role as yet unspoken of (as far as I know) and it's this: Anti Submarine Warfare.

Let me elaborate....the Astute class submarine now in use with the Royal Navy has been called the greatest current threat to the Russian Navy. Indeed, in the 2012 Exercise Fellowship, HMS Astute played a cat and mouse game for two days with the Virginia class USS New Mexico and left the Americans very impressed by its capabitilities. The problem is that these boats cost around £1.5 Billion each. The Royal Navy's traditional role in NATO has been to provide around 75% of the ASW assets in the North Atlantic but rising costs makes that more and more difficult to achieve.

But imagine a drone with tilt props akin to the V22 Osprey that can patrol the Atlantic for 20 plus hours at a time. Fit it with dipping sonar, arm it with Stingray torpedoes and you may have a very effective (yet relatively inexpensive) ASW craft that, because of its mobility and staying power, would be difficult for the submerged Russian subs to predict its whereabouts. Pilot fatigue could be erradicated by simply switching chairs in an office. Indeed, nationality could even be changed by switching control from the the USA to the UK in a matter of seconds.

Finally, with the continuing development of drone in-flight replenishment, these ASW drones could stay aloft almost indefinitely.

Hmmm. Maybe I should give the MOD or Pentagon a call

 

[Onyx]

But imagine a drone with tilt props akin to the V22 Osprey that can patrol the Atlantic for 20 plus hours at a time. Fit it with dipping sonar, arm it with Stingray torpedoes and you may have a very effective (yet relatively inexpensive) ASW craft that, because of its mobility and staying power, would be difficult for the submerged Russian subs to predict its whereabouts.

I think it is relatively more likely that the sonar units will be smaller drones themselves rather than a dipping buoy. Flying buoys would allow more triangulation and allow the mother craft to be a cheaper non-VTOL design.

I think that military stuff is likely to be more and more disposable as electronics and batteries get smaller.

 

[Foxbat]

I think it is relatively more likely that the sonar units will be smaller drones themselves rather than a dipping buoy. Flying buoys would allow more triangulation and allow the mother craft to be a cheaper non-VTOL design.

I think that military stuff is likely to be more and more disposable as electronics and batteries get smaller.

I see your point but I think the craft would have to be large enough to carry the weaponry able to take out a sub.

 

[Justin Swanton]

Indeed, modern aircraft are heavily computerized - but I presume the human pilot is there for a good reason.

I'm not railing against unmanned technology - as pointed out, we've had it for some time, whether in drones or even driverless cars. It just makes me uncomfortable the more technology takes away direct human control.

(One day driverless electric cars will probably be the norm, and I'll be one of those old fuddy-duddies refusing to get in one. )

Commercial jets have had autopilots for decades but an autopilot has never been a complete substitute for a pilot, who can respond to situations not forseen by the autopilot (and there will always be situations not forseen by an autopilot). It's the same for cars. Aircraft autopilots have a very simple journey to navigate: take off from a runway, fly whilst missing other aircraft (whose flight paths are known) and the ground, and land. Travelling in a car is far more complex: all sorts of things to miss, other vehicles whose behaviour you don't know, and a host of unexpected situations. I can't see a car autopilot ever allowing the driver to keep his hands far from the wheel. I certainly would never use one except to travel along a freeway on a quiet day.

 

[Justin Swanton]

I suspect the reason people find unmanned military drones creepy is because one imagines that since there isn't a human in them they are in danger of becoming self-aware and deciding the human race is a liability.

Which gets back to a certain thread on computers becoming intelligent like humans, but that's another story.

[just to add that nobody has anything to worry about]

 

[Vertigo]

Commercial jets have had autopilots for decades but an autopilot has never been a complete substitute for a pilot, who can respond to situations not forseen by the autopilot (and there will always be situations not forseen by an autopilot). It's the same for cars. Aircraft autopilots have a very simple journey to navigate: take off from a runway, fly whilst missing other aircraft (whose flight paths are known) and the ground, and land. Travelling in a car is far more complex: all sorts of things to miss, other vehicles whose behaviour you don't know, and a host of unexpected situations. I can't see a car autopilot ever allowing the driver to keep his hands far from the wheel. I certainly would never use one except to travel along a freeway on a quiet day.

There is one huge difference for cars; they can simply stop if the computer cannot cope with the situation and if there's not enough time to stop then it's even less likely that a human could react and take over in time to improve the situation. I don't know your age but I think the younger folk around today can expect fully automated cars within their lifetimes (they are incredibly close now), and I'd further predict that also within their lifetime they will probably not be allowed to take manual control within urban areas.

 

[-K2-]

I mentioned this earlier, though bears repeating again regarding Artificial Intelligence in autonomous vehicles, specifically aircraft (ground based is essentially easier (working in 2-dimensions) except for the collision avoidance and surprise conditions as noted above).

Though work began long before computer games, flight simulation software generated by gaming companies, for a couple decades has been seriously looked at by the military, even solicited to the point where new divisions have been added to those companies. IL2-Sturmovik by 1C-Maddox was one of those. The software used to reflect actual conditions (weather, temperature, altitude, air density, g-forces, etc.) were just a small fraction of the conditions considered. Changes to the aircraft in-flight, fuel use, ordnance expenditure, damage and so on, all taken into account.

They did so to give the game player a 'simulated' experience as close to real as possible (It's my understanding that actual pilots utilized MS-FS to practice difficult landings at difficult fields), and in kind, give AI aircraft realistic responses... EVEN, 'evolving/learning' tactics.

I'm not speaking of A to B - drop ordnance X - B to A, yet actually coping with threats from ground or air and then neutralizing them-- In 'real-time' as they come up. In that aircraft are now fly-by-wire, and the sensors can feedback information regarding the aircraft/weather/location/altitude state instantaneously, at the aircraft, not home base... It takes little to input that same simulation software into an actual aircraft and have it perform and REACT as though piloted.

And that is just what they make public

K2

 

[Onyx]

I see your point but I think the craft would have to be large enough to carry the weaponry able to take out a sub.

The mothership certainly could, and it doesn't need to be able to hover to do so.

 

[Onyx]

Commercial jets have had autopilots for decades but an autopilot has never been a complete substitute for a pilot, who can respond to situations not forseen by the autopilot (and there will always be situations not forseen by an autopilot). It's the same for cars. Aircraft autopilots have a very simple journey to navigate: take off from a runway, fly whilst missing other aircraft (whose flight paths are known) and the ground, and land. Travelling in a car is far more complex: all sorts of things to miss, other vehicles whose behaviour you don't know, and a host of unexpected situations. I can't see a car autopilot ever allowing the driver to keep his hands far from the wheel. I certainly would never use one except to travel along a freeway on a quiet day.

That seems like a bet you could lose within 10 years.

As far as unforeseen issues on airline flights, you could simply have remote emergency crews that can take over flying the plane if the automatic systems encounter something they are not able to cope with. Given instrument flight conditions and fly by wire controls, there is little difference between an in-aircraft crew and one in a simulator on the ground.

Except the radiation, of course.

I suspect the reason people find unmanned military drones creepy is because one imagines that since there isn't a human in them they are in danger of becoming self-aware and deciding the human race is a liability.

I think people have a much simpler concern about a machine not reacting properly to the situation, rather than worry that it will become alive.

People also tend to have more petty concerns - who will be held responsible if there is an accident.

 

[Lew Rockwell Fan]

. . . the first unmanned aeroplanes appeared as early as WWI - which surprised me: History of unmanned aerial vehicles - Wikipedia

Much earlier actually. Unmanned airplanes came before piloted ones. Samuel Pierpont Langley was building them in the 19th century.

 

[Lew Rockwell Fan]

I suspect the reason people find unmanned military drones creepy is because one imagines that since there isn't a human in them they are in danger of becoming self-aware and deciding the human race is a liability.

 

[Onyx]

Much earlier actually. Unmanned airplanes came before piloted ones. Samuel Pierpont Langley was building them in the 19th century.

I think the term "unmanned" in this context refers to aircraft having a control system that takes the role of the pilot.

Even in his attempts at manned flight, Langley didn't have a flight control system, making his airplanes the equivalent of heavier-than-air balloons.

 

[Edward M. Grant]

There is one huge difference for cars; they can simply stop if the computer cannot cope with the situation and if there's not enough time to stop then it's even less likely that a human could react and take over in time to improve the situation.

And then someone crashes into the back of them because they didn't expect to find a stationary car on the highway.

There's a much bigger difference between aircraft autopilot and car autopilot. In an aircraft, when something happens that the autopilot can't handle, there's typically several minutes for the crew to respond. In a car, there's likely to be a few seconds at most.

It's simply unrealistic to expect a driver to take over from a car autopilot when they're a few seconds from disaster and have spent the last three hours sleeping and watching pr0n.

Automated cruise control makes sense, because the driver knows it may fail and has to keep their eyes on the road. Fully automated driving with no manual controls makes sense, because the car has to be able to handle anything that might happen. Nothing in between makes sense in a world where trained airline pilots can fly a perfectly good airliner into the sea after the autopilot cuts out, when they have ten minutes to prevent it.

 

[Edward M. Grant]

As far as unforeseen issues on airline flights, you could simply have remote emergency crews that can take over flying the plane if the automatic systems encounter something they are not able to cope with.

An aircraft that can be remotely piloted is an aircraft that can be remotely hijacked.

Besides which, I was watching a video the other day recreating an emergency landing where an airliner lost all hydraulics and the only control the crew had was by throttling the engines to turn, climb and descend. They got the plane down and saved most of the passengers, but, as I understand it, no-one has ever managed to repeat the performance in a high-quality simulator. Knowing your life is on the line tends to concentrate the brain, and a remote pilot isn't going to have quite the same motivation to get it down safely at any cost.

 

[Onyx]

An aircraft that can be remotely piloted is an aircraft that can be remotely hijacked.

Unless the remotely piloted aircraft has a facebook account and uses its birthday as a passcode, it is utterly unlikely to be remotely hijacked. Do you know of any remotely controlled aircraft, satellites or spacecraft that have ever been taken over in the 100 years since their invention and use?

Besides which, I was watching a video the other day recreating an emergency landing where an airliner lost all hydraulics and the only control the crew had was by throttling the engines to turn, climb and descend. They got the plane down and saved most of the passengers, but, as I understand it, no-one has ever managed to repeat the performance in a high-quality simulator. Knowing your life is on the line tends to concentrate the brain, and a remote pilot isn't going to have quite the same motivation to get it down safely at any cost.

The simulator is likely the problem with this example. Doing close contact VFR work in simulators is extremely difficult because of a lack of detailed ground texture to judge speed and simulator lag - the inability of the simulator to match flight control inputs in real time - making 'seat of the pants' flight hard to emulate.

There are flight control programs developed by NASA that are very adept at controlling aircraft with missing or damaged control and lift surfaces. If a passenger plane was in extremis, I would hope an automatic system in the plane was making the landing and not a remote manual approach from a simulator operator. Machines can actually "sense" altitude, airspeed, inertial, angle of attack and attitude in a way a human pilot can only refer to, while a remote pilot's vestibular inputs are compromised by lag.


So my comment was basically that the aircraft could be remotely piloted in the case of an AI failure, but the AI would be the best pilot in the face of a mechanical failure. The likelihood of a combination AI and severe mechanical failure would be extremely low as those are not related systems.

 

[Edward M. Grant]

Unless the remotely piloted aircraft has a facebook account and uses its birthday as a passcode, it is utterly unlikely to be remotely hijacked. Do you know of any remotely controlled aircraft, satellites or spacecraft that have ever been taken over in the 100 years since their invention and use?

Everyone thinks their security is unbreakable... until it's broken. And, yes, Iran captured a US drone (possibly several?) a few years ago by taking it over in flight.

All it will take is for the bad guys to get one person into the infrastructure that controls encryption between the aircraft and the ground, and you can have a thousand 9/11s on the same day.

So my comment was basically that the aircraft could be remotely piloted in the case of an AI failure, but the AI would be the best pilot in the face of a mechanical failure. The likelihood of a combination AI and severe mechanical failure would be extremely low as those are not related systems.

Except you can only verify that the AI can handle situations you test it in. The crew apparently called the ground after they stabilized the aircraft, to get the procedures for flying their aircraft after a complete hydraulic failure... only to be told that there were no procedures, because a complete hydraulic failure was 'impossible'.

 

[Onyx]

Except you can only verify that the AI can handle situations you test it in. The crew apparently called the ground after they stabilized the aircraft, to get the procedures for flying their aircraft after a complete hydraulic failure... only to be told that there were no procedures, because a complete hydraulic failure was 'impossible'.

Of course a complete hydraulic failure is possible. What was likely thought "impossible" was that the crew were still alive to call for procedures because complete hydraulic failures rarely leave the aircraft in level flight attitude.

I just don't agree that an AI tasked to control the plane by whatever channels it has available is going to be unable to accomplish a differential thrust landing that a person could. An AI is more capable at that sort of adaptive piloting.

What an AI might not be able to do is something really out of the box, like the fictional upside down flight shown in the film "Flight", it should be able to very quickly decide what flight control it does have and calculate the effect.

 

[Foxbat]

Some news: US Navy awards Boeing $805.3 million contract to design, build MQ-25A Stingray

Just an observation....it's interesting when you listen to or read the views of test pilots nowadays. It seems that the software does most of the flying already (including target acquisition/priority selection, take-off and landing...vertical or otherwise). Seems to me that most advanced fighters are near-UAVs already.

 

[Vertigo]

Some news: US Navy awards Boeing $805.3 million contract to design, build MQ-25A Stingray

Just an observation....it's interesting when you listen to or read the views of test pilots nowadays. It seems that the software does most of the flying already (including target acquisition/priority selection, take-off and landing...vertical or otherwise). Seems to me that most advanced fighters are near-UAVs already.

I seem to remember being pretty much shot down in flames for making a similar observation earlier in the thread! In fairness I was making it generally whilst it's probably more true when referring to just high performance military aircraft.