Single pilot A350s, [i]There[/i]

[3WE]

Yes, it’s BEEN being talked about
IS being talked about
and WILL BE being talked about.

Evan and bean counters drool.

3BS Acknowledges that Cape Airways does it just fine (and this bizjet and that jet fighter) and considers it interesting, but I just don’t see it happening for several reasons…

One reason is that pilots are probably among the cheaper safety items, even if they are egotistical pains in the rear…

[supersean]

I can see this happening sooner on short haul flights but we will see democrats elect Trump as their party chairman before we see jumbos flown with only one pilot on long hauls

[Gabriel]

With one pilot aboard it will not be soon, not even for short flights (in big-ish planes).

The situation with long flights is that above 8 hours you need a 3rd pilot and above 12 hours you need a 4th. Only 2 of those pilots are in the flight deck at a time, the rest are resting, they rotate to meet the duty / rest periods. What Airbus is trying to do is to keep only 2 pilots, one of them in the deck and the other resting. So you will still have 2 pilots abroad. I think Airbus will manage to get this implemented within a number of years that can be counted with the fingers of a single hand.

[Not_Karl]

Half the stupid fatigues cowmonkey pilots in the cockpit, half the chances of them doing something stupid. Makes sense.

[3WE]

What Airbus is trying to do is to keep only 2 pilots, one of them in the deck and the other resting. So you will still have 2 pilots abroad. I think Airbus will manage to get this implemented within a number of years that can be counted with the fingers of a single hand.

Expletive!

I’ll repeat that guys die at the controls from time to time…

And if something bad happens that involves turbulence or nasty attitudes, or time critical things happen…not good. Can the resting pilot get there on time, or get there at all?

Repeating- the pilots might be relatively cheap!

I concede, though- it MAY happen, and automation is reducing pilot work load for when everything is hunky dory. One person can handle that…the bean counters may win.

[elaw]

Hmm... pilot flying + pilot monitoring, or pilot flying + nobody monitoring. I think I know which I'd prefer...

[3WE]

Hmm... pilot flying + pilot monitoring, or pilot flying + nobody monitoring. I think I know which I'd prefer...

But you are not_an MBA nor a bean counter. You have a bad attitude. Cut costs, increase income, Wall Street happy. Also, be sure to conspicuously signal virtues.

[Gabriel]

Hmm... pilot flying + pilot monitoring, or pilot flying + nobody monitoring. I think I know which I'd prefer...

It'd me more like HAL flying + Dave monitoring.

[ocelot]

Against my better judgement I looked there and found Gabriel, who should know_better, touting the virtues of computers as pilots...

Computers don't get task-saturated.

Sure they do. Not only that, they get it worse than humans -- typically when a human hits task saturation some things get done and not others, the quality of decision making suffers, obvious cues are missed... but generally with computers the result is that nothing gets done at all, or gets done so slowly that you might as well have not bothered. And for various reasons this has a tendency to strike in abnormal situations where one can least afford it. For example, in the 2003 northeast US blackout, during the event the utility's monitoring system fell hours behind real time.

(I could write more but there's no point going through in gory detail)

[Gabriel]

Airplane computers that are critical to safety are designed with enough capacity and bandwidth compared with the worst-case-scenario of data input/output and computing requirements... that they don't get task-saturated. Also they are designed exclusively for certain task, so you don't need an overarching OS than needs to be able to handle MSFS, zoom, chess, blender, afterFX, notepad, and things that didn't even exist when the OS was created.

[ocelot]

So are powerplant monitoring systems. In theory.

[3WE]

***In theory***

Indeed.

Excrement transpires.

[elaw]

An even-more-geeky-than-me coworker has a saying: "In theory, theory and practice are the same. In practice, they are different."

When I've thought about this in the past, one of the things I thought were against computers flying airplanes is they lack the intelligence/creativity of a human to deal with unforeseen situations. But in 100+ years of aviation... we've seen a he** of a lot of situations. At this point I'm inclined to think that if (and this is a big "if") we managed to program a computer to deal with all the weird things that aeroplanies have encountered in all the years we've been flying, as well as the non-weird things, the chance of "Hal" dealing with any given situation correctly should be just as high as a human for normal situations, and probably higher for non-normal ones.

But... that depends on programming the computer properly for a LOT of different scenarios. Some of which are fairly simple, but many of which are not. And computer programming even now often is not an exact science. That to me is worrisome.

[3WE]

An even-more-geeky-than-me coworker has a saying: "In theory, theory and practice are the same. In practice, they are different."

That is beautiful and calls for a bolded indeed.

[Gabriel]

An even-more-geeky-than-me coworker has a saying: "In theory, theory and practice are the same. In practice, they are different."

When I've thought about this in the past, one of the things I thought were against computers flying airplanes is they lack the intelligence/creativity of a human to deal with unforeseen situations. But in 100+ years of aviation... we've seen a he** of a lot of situations. At this point I'm inclined to think that if (and this is a big "if") we managed to program a computer to deal with all the weird things that aeroplanies have encountered in all the years we've been flying, as well as the non-weird things, the chance of "Hal" dealing with any given situation correctly should be just as high as a human for normal situations, and probably higher for non-normal ones.

But... that depends on programming the computer properly for a LOT of different scenarios. Some of which are fairly simple, but many of which are not. And computer programming even now often is not an exact science. That to me is worrisome.

I don't fully agree. I think that having computers flying autonomous planes (computers and planes specifically designed with "no pilot" as guiding input, not today computers in today planes) will be accompanied by an uptick of certain type of accidents that could have been avoided with a human on board. I call these situations "MacGyver". A Siux City case before Sioux City (now you can program that), a Hudson miracle or, even more, a TACA. There was a case with a plane that one side of the elevator got stuck in the full up position and the pilot avoided crashing first by keeping a high bank angle to keep the plane from pitching up, and then moving more pax to the FWD of the plane to be able to land. However, the avoidance of accidents where the human performance was a causal factor (Mr Bean situations) will go to zero. And that will grossly outweigh MacGyver.

I think that, pretty much like what happens with autonomous cars, the biggest barriers are not technical, but psychological (many persons say "I would never fly in a plane with no pilot") and legal/liability. When you have a computer screwing up in a system where that scenario is foreseen and the solution is tell Charlie to take over, if the plane crashes then it's Charlie's fault. In a case where the failure of the computer means the loss of plane and people, airplane companies will face lawsuits for the hundreds of lives they caused with zero recognition for the thousands they saved. And with that, regulations will be hard to change (regulators being political institutions).

From the technical perspective, I think we already have all we need to create an autonomous plane. Just that nobody put the pieces together yet.

[elaw]

You make some good points!

I agree 100% with the psychology aspect. People often don't trust computers, and with good reason. Even places with the best software QC (NASA for example) sometimes have bugs slip through. And the average quality of software is far from "the best".

As far as aircraft-flying computers meeting situations they don't know how to deal with, I agree that what you describe is possible, but I don't think it's inevitable. I think that could be avoided if a comprehensive effort were made to make the computer able to deal with: a) every known "normal" condition, b) every "abnormal" condition that's ever been seen (like the middle engine on a trijet exploding and taking out all hydraulics), and c) every never-before-seen-but-possible condition that could be envisioned. I'm talking filling a room with smart people, then having someone (preferably Jeremy Clarkson) say "What could possibly go wrong?", then devising a way to handle every possible response unless it's deemed impossible to deal with (wing-mounted engine explodes, engine part penetrates fuel tank, tank explodes and takes off half the wing).

I think if that were done, we could reach the point where when accidents occur, in each case (or at least the vast majority) when an accident does occur, it would be determined that the accident would have been unavoidable even if human pilots were present. The questions then become whether such an effort would be financially feasible, and would anyone bother to do it.

[3WE]

From the technical perspective, I think we already have all we need to create an autonomous plane. Just that nobody put the pieces together yet.

It's way past that. Drones have been doing this for quite some time, and not that "no one has put the pieces together", but no one has had the balls to do it. (NOT SAYING that it SHOULD be done.)

Now, all this being said, we (no italics) cannot handle science and technology worth a damn...Masks are useless, no wear a mask. 2-week lockdown to flatten the curve. Close Bob's shirt shop, but Wally World is open. Close fishing lakes in central IL, but let international flights land at ORD. Get vaccinated, whoops, let's mask up again...

And, to hell with pilots, what about ATC...not sure why we don't have a highly automated air-traffic system giving much more direct routes and tweaking speeds without someone clearing someone to the right runway, when it should have been the left, AND GIVING ORDERS WITH BOTH VOICE AND TEXT, WITHOUT STEPPING ON, GARBLING, MISUNDERSTANDINGS, ETC.

Nevertheless, back to my point- we could put two pilots up front and pay them regional pay rates and maybe do away with full motion simulator for recurrent training (just do it all with zoom meetings and MSFS)...that will help the public when HAL inevitably runs a perfectly good airplane full of souls into the ground. Heck, how would HAL have handled the MCAS crashes...a number of crews dealt with a phugoided MCAS pretty well.

[Gabriel]

As far as aircraft-flying computers meeting situations they don't know how to deal with, I agree that what you describe is possible, but I don't think it's inevitable. I think that could be avoided if a comprehensive effort were made to make the computer able to deal with: a) every known "normal" condition, b) every "abnormal" condition that's ever been seen (like the middle engine on a trijet exploding and taking out all hydraulics), and c) every never-before-seen-but-possible condition that could be envisioned. I'm talking filling a room with smart people, then having someone (preferably Jeremy Clarkson) say "What could possibly go wrong?", then devising a way to handle every possible response unless it's deemed impossible to deal with (wing-mounted engine explodes, engine part penetrates fuel tank, tank explodes and takes off half the wing).

I think if that were done, we could reach the point where when accidents occur, in each case (or at least the vast majority) when an accident does occur, it would be determined that the accident would have been unavoidable even if human pilots were present.

Well, in some instances, accidents were avoided by humans in cases where an accident was (almost) unavoidable with human pilots present.
Take for example United 2323. Many crews attempted to replicate the she situation in the sim and ALL of them failed to control the plane.
Or take TACA 110 that landed in a levee with both engines failed. They had committed to land in the water and at the last moment the FO saw the levee and suggested it to the captain, who then did a side-step to the levee and because he was too high he did a sideslip and finally finely adjusted its trajectory laterally to negotiate the space with obstacles left and right. I don't see a computer doing that anytime soon until we have an actual, real AI that is not just a neural network or machine learning.
Also the example I mentioned before of the captain summoning the pax to the front to move the CG.
It will be very hard, with today's technology, to have a "creative" computer that can devise a solution using all the resources available, including those that were not supposed to be "resources available".

A computer will be better than any human pilot to glide and adjust the descent path to safely land a plane that lost all the power, at an airport in their database. Now, visually acquiring target places for off-field landings, deciding which one is the best, and taking last-second avoidance maneuvers (trees, power lines)... I don't see that happening with the current technology.

Now, that should not be a limiting factor. If you have a system that could avoid 95% of the fatalities while creating 5% "new" fatalities in the process, that is still a 90% of life saving. The problem is that that 5% not saved and especially the 5% "new" ones will get 99.9% of the attention and the 90% saved will get 0.01% of the attention. As it is happening now with self-driving cars, "Oh look, the idiot car didn't realize that there was a truck lying on its side across the highway, stupid car, sue Elon Musk...". Now (this is news) the NIHS has opened an investigation into 13 accidents involving Teslas not correctly identifying emergency vehicles on the side of the road marked with flashing lights, arrows or cones, one fo which resulted in one death. Don't get me wrong, go for it, that's how we improve safety. But no one is taliking about how many accidents and lives the Tesla saved.

[Gabriel]

It's way past that. Drones have been doing this for quite some time, and not that "no one has put the pieces together", but no one has had the balls to do it. (NOT SAYING that it SHOULD be done.)

Drones don't need or have the level of fail-safe-ness required by passenger-carrying planes.
Losing the vehicle is a more acceptable outcome when you have 0 pilots and 0 passengers vs losing the vehicle + 0 pilots + 400 passengers.

Now, all this being said, we (no italics) cannot handle science and technology worth a damn...Masks are useless, no wear a mask. 2-week lockdown to flatten the curve. Close Bob's shirt shop, but Wally World is open. Close fishing lakes in central IL, but let international flights land at ORD. Get vaccinated, whoops, let's mask up again...

Very apples to oranges. COVID is a dynamic situation for which our understanding and power to model the future and the what-ifs is very limited at best. It would be like a computer flight-testing a revolutionary aircraft design,

Heck, how would HAL have handled the MCAS crashes...a number of crews dealt with a phugoided MCAS pretty well.

the MCAS crashed would have never happened in an autonomous plane not because the computer would have successfully recovered, but because the autonomous plane would have several sources of data, preferably at least one of them independent from the others (based on a different principle), so a single-sensor failure would not have been a critical problem, Now, if it gets to the point where the remedy is "kill the electric trim and turn the trim wheel manually powered by muscle", that would have been a problem with no muscle available in the cockpit to turn the trim wheel manually. Besides, if you install a technological muscle to turn the wheel "manually" when commanded by the computer, the computer would have never commanded that since it was the computer itself who was commanding "we need more trim down, more trim down, MORE TRIM DOOOOOWWWWWNNNNN!!!". Of course, the computer would have been programmed different and, again, the MCAS upset would have not even happened.

[3WE]

Very apples to oranges. COVID is a dynamic situation for which our understanding and power to model the future and the what-ifs is very limited at best…

I think you misunderstand- THE GENERAL PUBLIC’S ACCEPTANCE of pilotless planes is likely to be driven emotion instead of science.

While biology (and CoVid masks and vaccines) are nowhere cut and dry as composites and computers (and we are in a dynamic state of learning), THE GENERAL PUBLIC’S RESPONSE to it all is often driven by emotion instead of science.

Not identical, but similarities…

[Not_Karl]

How about this: Pilot-less aeroplanie, but at least one passenger has to be profficient in the latest version of MSFS.

[Gabriel]

How about this: Pilot-less aeroplanie, but at least one passenger has to be profficient in the latest version of MSFS.

Why the latest?

[3WE]

How about this: Pilot-less aeroplanie, but at least one passenger has to be profficient in the latest version of MSFS.

…and it’s a 50-dollar upgrade to sit in the cockpit.

Brill-yunt!

[elaw]

How about this: Pilot-less aeroplanie, but at least one passenger has to be profficient in the latest version of MSFS.

Why the latest?

Because if you can deal with the bugs in that program, you can deal with anything an airplane throws at you?

Imagine you're flying along fat, dumb, and happy with autopilot on (but no FMS). You need to make a heading change, so you start twisting the heading knob on the AP. After you've turned it a random number of "clicks" it starts spinning by itself at about a 180°/second rate (with the AP trying to follow) and all the controls become unresponsive.

[3WE]

...so, we can’t even get fake aeroplanies to act right, but we want computers running the real ones...

Not like MSFS has had a LOT of programming and debugging over a lot of years...