Surely you can't be serious: Airbus close to landing fully automated passenger jets

UpNext, Airbus's future technology-focused subsidiary, reported on Thursday that it has entered the final three months of testing tech it hopes will automate the process of getting a plane from the air to the gate. The system, called DragonFly, tackles automated operations like diversions, landings, and taxi procedures through …

[abetancort]

An step in the path to reduce the number of pilots needed to fly commercial.

I think that airbus is aiming at allowing its new planes to be certified to be operated with just one pilot onboard. If the plane advanced autopilot allows the pilot to do the checklist by himself and in case of an emergency takes over the plane safely, it will allow the pilot to concentrate in solving the emergency (engine fire or malfunction) or if the sole pilot is disabled it’s able to safely land the airplane in the nearest airport, it will allow the airplane to be safely flown with one just pilot reducing the crew cost of operating the plane.

[Pascal Monett]

Re: reducing the crew cost of operating the plane

On paper, yeah. So, for the beancounters, as usual.

How do you tell the autopilot that there's a crosswind ?

And how does it manage windshear ?

[Little Mouse]

Re: reducing the crew cost of operating the plane

Just rename the Autopilot to "Full Self Driving Beta", to manage expectations...?

[FrogsAndChips]

Re: reducing the crew cost of operating the plane

Shirley it should be named Otto?

[spireite]

Re: reducing the crew cost of operating the plane

If it means the pilot(s) can relax, then surely it should be called Ottoman.

[Kev99]

Re: reducing the crew cost of operating the plane

I wonder if Tesla is involved in the design?

[M.V. Lipvig]

Re: reducing the crew cost of operating the plane

Oh great, now we'll have airplanes running into small children and emergency vehicles.

[An_Old_Dog]

Re: reducing the crew cost of operating the plane

How well does this computerized thing work when birdshit covers one or more of the sensors? Or when one or more birds strike the nose of the plane, where many of the sensors are?

Having multiple pilots in the cockpit isn't just about reducing workload. It's about having in-air insurance in case things go wrong, and having multiple knowledgable, trained human minds working on the problem, in case one of them freezes, or has a brain-fart. You know, "redundant systems."

Just as you don't "need" automobile insurance until you've had an auto-crash, you don't "need" multiple pilots until things go pear-shaped.

I don't trust a bunch of random programmers -- no matter how many of them are Ph.ds -- to think of every possible in-air problem scenario and to construct the correct response to each of them.

[Anonymous Coward]

Re: reducing the crew cost of operating the plane

s/random programmers/pilots/

Nobody can think of every possible in-air problem scenario and construct the correct response to each of them.

[Anonymous Coward]

Re: reducing the crew cost of operating the plane

so Adam Nobody switched from protests to programming?

[Anonymous Coward]

Re: reducing the crew cost of operating the plane

Yeah but I'd 100% rather have someone up front who's trained and has got skin in the game when it comes to taking a punt at resolving unpredicted scenarios, rather than a bunch of people in a warm office on the ground standing around a whiteboard, or some half-arsed attempt at AI. I'd be happy to pay more to fly knowing there are two people up there and no dodgy AI, than one person and an AI that might decide it knows better.

And I know there have been pilot murder-suicides - but I suspect someone really determined would find a way to turn off or overpower the AI, or do something the AI won't be able to recover from. For example, if you fire all the fire extinguisher bottles into all of your engines, you're not restarting them, and nor's the AI; do it far enough away from a suitable airport and no AI's going to save you...

[Naich]

Re: reducing the crew cost of operating the plane

You should get in touch with Airbus. I'm sure they haven't thought of those things.

[First Light]

Re: reducing the crew cost of operating the plane

Shocked at all the downvotes. Your questions are entirely reasonable.

[Anonymous Coward]

Re: reducing the crew cost of operating the plane

They're reasonable questions if someone is asking to gain knowledge and understanding.

They're unreasonable if the one asking assumes that they're much smarter than the people who developed the system.

On the Internet, it can be hard to tell which situation exists.

[Anonymous Coward]

"On the Internet, it can be hard to tell which situation exists."

...and if in doubt about the motivation behind a question (as opposed to its actual content), we give a downvote, just to be on the safe side?

As the old pre-Internet ditty has it:

If the reasons of this clown

do not please or are not known,

let's just simply vote him down.

:-)

[Anonymous Coward]

Re: reducing the crew cost of operating the plane

Ohh - a hard one. maybe get satellites to send out precise time signals so that you can do some clever maths and work out a real-time location and velocity vector in 3 dimensions, and then compare that to some sort of system to measure airflow?

[Anonymous Coward]

Re: reducing the crew cost of operating the plane

These little draughts can be pretty localised and peaky, so I wish you luck inventing a system for that that covers the whole downward trajectory. It's exacty because they're non-stable that they are a problem.

[NerryTutkins]

Re: reducing the crew cost of operating the plane

Do real pilots deal with them? How?

I am pretty sure that however it is that real pilots deal with them (e.g. height drops quickly) could be dealt with by automation too.

[Kev99]

Re: reducing the crew cost of operating the plane

With years of experience, help from both pilots, and guts.

[hoola]

Re: reducing the crew cost of operating the plane

The accuracy with altitude on GPS is not great. In reality it will have to be based on some sort of digital altimeter.

[Jan 0]

Re: reducing the crew cost of operating the plane

> it will have to be based on some sort of digital altimeter

I'd expect it to be based on radar and an accurate, digital contour map.

[Anonymous Coward]

Re: reducing the crew cost of operating the plane

Having seen what crosswinds can do I fully agree with the question - there are some truly scary landings out there which demonstrate just what an experienced pilot can manage. I expect the first releases of this to probably declare cross and shear winds as an out of bounds situation for the software.

Let's just hope it doesn't do a Tesla FSD and hands back the plane half a second before it gets into trouble..

[NerryTutkins]

Re: reducing the crew cost of operating the plane

Most modern fighters are inherently unstable and can only be flown thanks to fly-by-wire. Machines have reactions and response times no human can match. I am pretty sure that suitable automation could handle crosswinds and wind shear way better than any human pilot could, if equipped with the right sensors and programming.

[Bubba Von Braun]

Re: reducing the crew cost of operating the plane

Airbus has been using Autoland for decades.. However this does need an ILS approach, three buttons and the Airbus will brake to a stop on the center-line of the runway.

Bigger issue is ground taxi, as with AutoDriving Beta's too many random items moving around between rampies, service vehicles and other aircraft.

Me, I want redundancy on the flight deck Autoland or not, last thing we need is to see a ground taxi goes across a runway and is hit by an Aircraft on a takeoff roll.

Or has life become so cheap, we are happy to consign 150-300+ people to death or injury to save a few bucks on an airline ticket?

BvB

[RPF]

Re: reducing the crew cost of operating the plane

GPS-only autoland is now possible. Regulators still being cagey about it, though.

[werdsmith]

Re: reducing the crew cost of operating the plane

In my day this kind of trickery was being tested with Wide Area Augmentation System (WAAS) for GPS. This is a ground based system that increases the usable accuracy of GPS. Autoland can only operate where the ground systems are in place and this is at major facilities that can justify it.

[Anonymous Coward]

Re: reducing the crew cost of operating the plane

Or has life become so cheap, we are happy to consign 150-300+ people to death or injury to save a few bucks on an airline ticket?

That's a question for Boeing..

[M.V. Lipvig]

Re: reducing the crew cost of operating the plane

Life is cheap to a bean counter. Remember, it was a bean counter at Ford that decided that the vslue of people who would burn to death in a Ford Pinto was less than the 7 bicks and a bit it would cost to prevent it.

[werdsmith]

Re: reducing the crew cost of operating the plane

Bigger issue is ground taxi, as with AutoDriving Beta's too many random items moving around between rampies, service vehicles and other aircraft.

This is not a big problem with single pilot operations. Nobody is suggesting pilotless yet.

but if you did, then I presume the aircraft would be towed out to the runway hold point. There are already some trials going on towing the aircraft out to the hold in order to save the taxi fuel.

[Jan 0]

Re: reducing the crew cost of operating the plane

> demonstrate just what an experienced pilot can manage

Don't the latest "unstable by design" military aircraft, show how well the avionics can handle the pilot's slow and sketchy input?

[UK DM]

Re: reducing the crew cost of operating the plane

Huh, crosswind and shear are easy things to manage. Detectable, weather quality is reported, corrected with flight control input reactions, all things already going on with such a system.

The article talks of the system managing the flight controls so the pilot can manage the critical incident with the aircraft.

But surely that incident has a high chance of causing the system to disconnect itself because it is not operating within the allowed envelope that has a better than 1 in 1 billion flying hours of failure, which I understand to be a benchmark all good aircraft manufacturers demonstrate they achieve for certification.

[Anonymous Coward]

Re: reducing the crew cost of operating the plane

Crosswinds I can see your point. FMCs already have METAR data and that can be used as the basis for what to expect. Windsheer though by definition is an unexpected and unpredicted sudden change in wind direction. Its possible that in the future autopilot will be able to deal with it, but this project is not trying to do so.

Airbus though are not claiming this will remove the need for pilots, they are saying that it can be used in good weather conditions to help pilots when they are task loaded with other issues (such as a bird strike) or a medical emergency in the cockpit.

As long as pilots don't rely on the system its another useful tool. And as pilots we love flying and are not about to put the autopilot on and rest our feet for the rest of our career just because the plane has more capabilities than before.

[werdsmith]

Re: reducing the crew cost of operating the plane

As a pilot, I feel windshear as an increase in sink rate, a reduction in G which I presume my inner ear detects, then I add more throttle, a hand stays on the throttle the whole time in readiness. I would say that a sensor could also do that. Some places I approach and there is never any windshear, it is geographic conditions that produce it. For instance, approaching Luton from the west where there is a kind of cliff edge before the runway. I alway expect windshear and I'm ready for it.

Crosswinds. Well, pilots have their limits and so do the aircraft have a max demonstrated crosswind component. If I am going to be outside this on my approach then I just have to divert to a more into wind runway somewhere else. Same if I can't land due to visibility minima.

Maybe airports need more tarmac so one runway is always into wind.

I find there are few things more satisfying than a greaser on the centre line after an approach with loads of crab or slip.

[David Hicklin]

Re: reducing the crew cost of operating the plane

> Maybe airports need more tarmac so one runway is always into wind.

Mount the runway on a big turntable ?

[nautica]

Re: reducing the crew cost of operating the plane

Crosswind? Wind-shear? To the bean-counters? All just minor details. A mere bag of shells, as Inspector Jacques Clouseau would say.

And quit calling me 'Shirley'.

[jgarbo]

Been doing it for years

Airbus's ATTOL project has been taxiing, taking off and landing planes for years. Most planes can land automatically if necessary. Relax.

[MajDom]

Re: reducing the crew cost of operating the plane

How do you tell the autopilot that there's a crosswind ?

And how does it manage windshear ?

How does a pilot learn how to do crosswind and windshear landings? Doing them a thousand times in a simulator?

How many times do you think a deep learning system is able to go through simulations and how long before it can handle them absolutely flawlessly? If it takes a million, it will do them, and I doubt they'll be worse at it than humans.

[Mishak]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

The one thing I'm not sure about is how a single pilot will cope with a major systems fault. It is easy enough to cope with a "simple" failure (such as the loss of a single engine), but it is very easy to get in a "human factors" mess if something unexpected happens (such as a failure that is considered so unlikely it is not taught in a simulator) - and that is where having a second pair of eyes, hands, ... is very useful.

It would not be fun having to run checklists at the same time as having to manually fly a wounded aircraft, especially when near to the ground.

[Richard 12]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

Even engine failure isn't "simple".

There have been several incidents where the overstressed pilot shut down the wrong engine, for example.

[Oh Matron!]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

I've been binge watching Mentour Pilot on Youtube, and what is VERY apparent is that, under high workloads, pilots panic and do the craziest things, including shutting down the wrong engines. Pretty much nearly every crash is due to operator error rather than equipment.

HOWEVER: There's an astoundingly big caveat to that: Some of the situations are because there's no president: The Air France flight from Brazil, for example: To Have all three speed readings wrong was unheard of, and the pilots had only trained for stalls at low level, not at 35,000 feet.

[Anonymous Coward]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

I misunderstood your comment, and thought that "no president" was some unusual variant of the "best of three" voting system where one could over-rule the other two. Then I realised you meant "precedent".

[Anonymous Coward]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

I blame that on auto-incorrect. I've had a few of those.

[Sceptic Tank]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

Check out 74gear as well.

[Gene Cash]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

There's an awesome blog called "Fear of landing: the art of not hitting the ground too hard" https://fearoflanding.com/ where she tackles air incidents from the perspective of a pilot.

It's really good. No connection other than being a mesmerized reader.

[Marty McFly]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

Mentour Pilot rocks! He recently did a video on single pilot commercial aircraft. A good fact based analysis. His conclusions...Maybe, and not for a few decades at the earliest.

[Lars]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

"Mentour Pilot rocks!"

Well I do agree but I must admit I find it painful to listen to his great effort to speak educated English in Swedish.

Also I think he is a Boeing pilot only.

This video, well made by him, is a good example of how two pilots make more silly mistakes than any automatic landing system would do, most likely.

https://www.youtube.com/watch?v=o-IuOnDBoGA

HOW could they let it GO THIS FAR?!

A very experienced pilot on both Boeing and Airbus is Juan Browne on www.youtube.com/@blancolirio

[RPF]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

A stall is a stall. Those AF guys messed up hugely, sad to say.

[MyffyW]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

I challenge any autopilot to be able to repeat this sequence:

Sully: "You got any ideas?"

Skiles: "Actually Not"

Sully: [Raises One Eyebrow]

Thirty seconds later, flawless landing on the Hudson

[A.P. Veening]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

As that flawless landing on the Hudson now is in the books, I am pretty sure Airbus took it into consideration for Dragonfly. After all, Airbus was smart enough to re-order the sequence of steps in the checklist (first action of Sully was to activate the APU in the tail, while that was on the bottom of the checklist, giving him power to control the plane).

[Mishak]

Saved my bacon

Simulator only (luckily), but I had a twin engine failure at 3000ft climbing out from Belfast - first action was "APU start" before thinking about the best way back to the runway. Worked so well I then had a serious problem bleeding energy before reaching the threshold.

[hoola]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

Ahh, but there you are wrong, the major failure will not happen as everything is double or triple redundancy, cross checked and so on.

All the things that smart people believe are failsafe until the event they did not think of occurs.

To be fair, physical reliability has improved hugely however the outcome now is that the event is either minor or catastrophic with very little in between. The scary part is that system reliability had generally got worse as complexity and complacency have increased.

[Lars]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

The first step is most likely to cut down to two pilots from three or four on long flights.

Which is one pilot at the helm and one resting flying, while leaving and arriving is with both pilots in the cockpit.

I cannot see a huge problem there.

Also one has to remember that a sane company cannot afford not to work towards those possibilities.

[RPF]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

Because "SAFETY is our first priority" (TM)?

[Anonymous Coward]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

I note the article says "designed to land and taxi when the pilot is incapacitated, or just busy"

What exactly could the pilot be doing, which is more important than landing the plane?

[hoola]

Re: An step in the path to reduce the number of pilots needed to fly commercial.

Texting or updating social media on their phone

Drivers appear to believe it is more important than being in control of their car all the time.

To be fair to the pilots one would hope that it is very unlikely they would be using a personal device "at work".

[Horridbloke]

Just one question...

Is it inflatable?

[seven of five]

Re: Just one question...

Only from the Stewardess.

[Graham Dawson]

Re: Just one question...

Surely you can't be serious.

[MaddMatt]

Re: Just one question...

I am serious, and don't call me Shirley.

[GlenP]

Re: Just one question...

Don't call me Shirley!

(Someone had to say it).

[Mayday]

Can be done - small boys doing it already

The Cirrus SF50 has an emergency landing system for the event of pilot incapacitation.

https://cirrusaircraft.com/story/cirrus-aircraft-earns-easa-approval-for-vision-jets-safe-return-emergency-autoland-system/

Unfortunately my crappy IT salary doesn’t put me in the market for a personal jet.

All that said, even if heavies are approved to fly with one/no pilot I’m not so sure the market (ie paying passengers) will go for an airline who has robots driving the thing.

[ChoHag]

Re: Can be done - small boys doing it already

It will be seen as a feature.

Also, cheap flights.

[Anonymous Coward]

Re: Also, cheap flights.

also, more profit (yes, I know, it's absolutly NOT the reason it's happening, nosir)

[Antron Argaiv]

Re: Can be done - small boys doing it already

No, silly person, the flights will NOT be cheaper. The cost will actually increase, while the seatroom will continue to decrease. You'll also find a credit card reader over the toilet roll.

[Anonymous Coward]

Re: Can be done - small boys doing it already

the market WILL go. This argument we've hand many times in the history of the so-called human civilisation. If something can be done, in order to make somebody richer (and to hell with passangers' last moments) - it WILL be done.

[dogcatcher]

Re: Can be done - small boys doing it already

I got worried when they dropped the rule that only four-engined aircraft could be used for transoceanic flights, that was after they had booted out the navigator and flight engineer. I'm staying home if they introduce a Tesla type system, especially if it is as short-lived as Dragonflies are, as they only fly in the summer and spend most of their lives at the bottom of a muddy pond.

[EvilDrSmith]

Re: Can be done - small boys doing it already

"I got worried when they dropped the rule that only four-engined aircraft could be used for transoceanic flights, that was after they had booted out the navigator and flight engineer."

A reasonable attitude to take at the time perhaps, but have those decisions proved to be wrong? I'm not aware of any issue that arose because a transatlantic flight was by a twin not a four-engine.

Not that I'll be rushing to board any self-drive passenger 'planes anytime soon, mind.

[Uncle Slacky]

Re: Can be done - small boys doing it already

You can thank increased engine reliability leading to ETOPS, aka "Engines Turn Or Passengers Swim".

[hoola]

Re: Can be done - small boys doing it already

I suspect that for most people the only thing the care about is the cost to get to their latest cheap holiday destination. How they get there and who is doing the driving/piloting is a total irrelevance.

On the other hand I know some people who are friends of a friend who never put both children on the same plane as both parents. The parents are always split, one with the children. one on a separate flight.

The lunacy of this becomes even better when they will quite happily all get into the same vehicle to drive 100 miles to get to the airport.

[A.P. Veening]

Re: Can be done - small boys doing it already

I’m not so sure the market (ie paying passengers) will go for an airline who has robots driving the thing.

They will start with cargo planes.

[Mayday]

Cargo planes

Interestingly enough, Australia has different aviation medical requirements for pilots of commercial (freight under 8168kg vs other commercial) and private pilots.

[RPF]

Re: Can be done - small boys doing it already

Totally different requirements for that type of aircraft.

[MyffyW]

Looks like I picked the wrong week...

Steve McCroskey : They could be miles off course!

Rex Kramer : That's impossible. They're on instruments!

[cuts to the cockpit, where Striker, Elaine, Randy and Rumack are all playing musical instruments]

[KittenHuffer]

Did you really need to ask that?

"The video does not, however, explain how the aircraft communicates with air traffic control for clearance into controlled airspace if the pilot is incapacitated, as that task is completed through human verbal interaction."

I would have thought that as soon as this system is required due to the pilot being incapacitated that ATC would drop everything and clear flights out of the way to allow this aircraft to land as a priority. I don't think ATC are going to be telling the aircraft to go into a holding pattern before entering their airspace.

[Richard 12]

Re: Did you really need to ask that?

Exactly.

This is only for a full-blown Mayday. ATC will be keeping everyone else out of the way.

As they already do, because a single pilot landing the aircraft is overworked and so needs the simplest flight path possible.

[A Non e-mouse]

Re: Did you really need to ask that?

I think humans pilots incapacitated and computer taking over falls into "Mayday" territory!

[Korev]

Re: Did you really need to ask that?

> I think humans pilots incapacitated and computer taking over falls into "Mayday" territory!

And what about the other 364 days of the year?

[A Non e-mouse]

Re: Did you really need to ask that?

My father used to work in ATC. He said that as soon as an aircraft declared an emergency, the skies would part to get that aircraft on the ground ASAP.

[Michael Hoffmann]

Re: Did you really need to ask that?

And the accident reports are full of case where anybody reading would go "why didn't they just declare an emergency and get the damn thing down?"

On the tombstones you could write "they didn't want to deal with the paperwork". (stole that from Rod Machado)

[Anonymous Coward]

Re: Did you really need to ask that?

I vaguely remember a couple of years ago ryanair had a great, great idea to 'optimise' how much fuel their aircraft hold, both to 'optimise' the cost of each flight, and to avoid (vicious rumour, no doubt) having to wait their turn on approach. I think they did get away with it more than once, declaring an emergency on account of low fuel, but then the story disappeared from media, my gut feeling is that 'somebody' had a _very_ quiet discussion with ryanair behind _very_ tightly closed door about what you can do about improving your bottom line, and about what you won't. But hey, if you don't try, you don't get...

[imanidiot]

Re: Did you really need to ask that?

I suspect Ryanair figured out multiple "Ryanair flight had to declare emergency for nearly running out of fuel on routine flight" news articles in short succession probably hurt it's bottom line and reputation more than the marginal gains in efficiency from carrying a few hundred pounds less fuel. People already know Ryanair is shit is most ways, but at least they seem to have an OK-ish safety record. How much of that is because they actually want to or because they're dragged kicking and screaming by the aviation authorities is anyone's guess.

[A Non e-mouse]

Re: Did you really need to ask that?

probably hurt it's [...] reputation more than the marginal gains in efficiency

I didn't think low cost carrier cared about their reputation. They treat their customers like s**t, the customers know they're going to be treated like s**t but nobody cares as they think they're paying £1 a flight*.

* Exlcuding all the optional fees that are required to be paid for you to board the aircraft.

[Anonymous Coward]

Re: Did you really need to ask that?

I think coming in low on fuel breaks actually a couple of safety rules. There's a reason they are required to fly with a safety margin.

Also, I suspect that declaring a mayday does come with a metric boatload of paperwork afterwards (certainly for commercial flying which risks many more souls* in one go), and as it concerns air safety it also involves a review which again could lead to consequences for what is clearly abuse of safety measures. You can get away with that once, the second time you'll face questions and the third time you may end up with a flight ban pending investigation.

* _{You can deduct bankers and lawyers from that total, they don't have one.}

[A.P. Veening]

Re: Did you really need to ask that?

Ryanair only had those low fuel problems on a couple of diversions, had the flights gone as planned, there would have been no problem. But they were bending the regulations on the amount of fuel necessary for diversion plus holding.

[The Oncoming Scorn]

Re: Did you really need to ask that?

Capt. Malcolm Reynolds: Well, just get us on the ground.

Hoban ‘Wash’ Washburn: That part’ll happen pretty definitely.

Icon - Browncoat.

[imanidiot]

Re: Did you really need to ask that?

But how would ATC know? If the pilot has time to squawk an emergency code (7700) then maybe, but otherwise Aviate, Navigate, Communicate. ATC will the last to know there is an emergency and if the pilot is knocked out before he can communicate they wouldn't know unless the plane itself has ways of ensuring that. I'd be leery of any system that could auto-squawk an emergency code though.

[A.P. Veening]

Re: Did you really need to ask that?

As a first guess I'd say Dragonfly will communicate to the cockpit it will take over in <appropriate, short duration> unless countermanded. Once it takes over, it is sure within its rights (and responsibilities) to squawk an emergency code and even more so if it waits a little bit more after taking over.

[Phones Sheridan]

Re: Did you really need to ask that?

The other thing would be cabin crew. They can't enter the cabin, but if they feel something is off*, they can knock on the door, no reply means they press the red button in the Kitchen, the alarm goes off in the cockpit, along with a countdown to the computer taking over. Easily cancelled if the pilots are responsive, no response and Otto takes over, calling in a mayday at the same time. And if an annoying terrorist or passenger keeps pushing the button, then since 9/11 he's going to be very quickly on the receiving end of some summary social justice meted out by greater number of other passengers.

*Cabin crew familiar with routes know when a descent, turn, delay is out of the norm.

[M.V. Lipvig]

Re: Did you really need to ask that?

"then since 9/11 he's going to be very quickly on the receiving end of some summary social justice meted out by greater number of other passengers."

That is the only silver lining to that whole mess. Hijackings are a thing of the past, and attempted hijackers don't have to wait on a mere trial and sentencing to receive their penalty. They'll be beat to death on the spot and 400 passengers will swear up and down that they saw the hijackers beat themselves to death.

[Anonymous Coward]

Re: Did you really need to ask that?

The Cirrus SF50 system referenced by another commentator handles declaring an emergency on appropriate frequencies. I don't know all the details, but sounds like it essentially says "I'm going to runway xx at airport yy due to an emergency, please get out of my way!"

[David Hicklin]

Re: Did you really need to ask that?

I guess the system would set an emergency transponder code

[Duncan Macdonald]

Automated emergency landing

There is already an automated emergency landing system available for light aircraft - Garmin Autoland. When this system is activated (by a guarded switch on the console) it takes full control of the aircraft - notifies ATC - selects the best available airport and lands the aircraft and stops on the runway with no further human input required. This system is intended to cover the pilot becoming incapacitated on an aircraft with a single pilot.

The current high end Autoland (CAT 111c) systems available for big jets require the pilot to select the runway but can safely land the aircraft even in zero visibility.

The first automatic landing of a commercial jet with passengers aboard was on 10 June 1965 - see the Autoland article on Wikipedia for more details.

[Gene Cash]

Re: Automated emergency landing

Only if you've got the eye-wateringly extremely expensive full Garmin glass cockpit installed.

Not only can most people not afford it, but there's a lot of light aircraft that just can't accommodate it.

[Phil O'Sophical]

Re: Automated emergency landing

I remember a 1980s BA shuttle flight operated by the new 757, where the landing was more than a little firm (not dangerously so, just unexpected). As we taxied the pilot came on the intercom and apologised for the hard touchdown, saying "sorry about the bump, it's the first time we've flown one of these and we thought we'd see how it managed on its own."

I was glad he waited until after the landing to tell us that.

[Lord Elpuss]

""A dragonfly has phenomenal vision, the ability to see in 360°, and can recognize landmarks, which in turn help it to define its territorial boundaries," explained Airbus."

Translated: We wanted to call it Dragonfly, and looked for any and every reason to try to make this make sense.

[Mr F&*king Grumpy]

At least they didn't try to manufacture a reverse acronym....

[Antron Argaiv]

The technical term is "backronym"...and someone is probably working on one right now

[Strahd Ivarius]

"someone"...

an Airbus guy or an el Reg commentard?

[PhilipN]

But could it land on the Hudson?

Or ..... "I don't like the look of that weather system" etc etc

[Richard 12]

Re: But could it land on the Hudson?

Better than an average commercial pilot, certainly. The existing autopilot and autoland is better at keeping the aircraft on the chosen flightpath, wings level, flare at the right moment than most pilots.

It would not be able to choose to do so, though. It'd panic and hand over to the pilot the moment it determined it could not reach a designated runway.

Fortunately it should reach that conclusion within a few seconds of the failure, perhaps giving the pilot more time to make a choice about what to attempt.

[Anonymous Coward]

Re: But could it land on the Hudson?

Fortunately it should reach that conclusion within a few seconds of the failure, perhaps giving the pilot more time to make a choice about what to attempt.

You're in Tesla FSD territory here - would a pilot really be able to maintain enough situational awareness if he/she had not been flying themselves up to that point to take over? Handing off when it gets too complicated may help to avoid liability for the auto landing system manufacturer, but it does not exactly strike me as an ideal point for the pilot to retake control.

I'd say the system should be at least smart enough to detect that things could get hairy and give the pilot a heads up that manual control may be required - that way it won't come as a complete surprise. Surprises are *bad*, planning ahead is good.

[GrahamRJ]

Re: But could it land on the Hudson?

Ironically, a fully automated system maybe wouldn't have had to, because it could have made the decision to divert to LaGuardia instantly, and flight simulators suggested it would have got there. By the time Sullenberger and Skiles had figured out WTF just happened, carried out all the manual steps to recover control, and then looked at what survivable options they had left, there wasn't another survivable option.

[Phil O'Sophical]

Re: But could it land on the Hudson?

The biggest problem in any automated disaster recovery setup, not just with aircraft, is that reliably determining WTF has happened is very difficult. Humans still tend to be better (but not necessarily faster) at it than machines, which are limited by the scenarios they've been programmed to expect.

[Anonymous Coward]

Re: But could it land on the Hudson?

I don't think so. Simulator tests only supported that theory until it was pointed out that pilots in the simulation already knew upfront what was going to happen.

In addition, we barely trust automatic pilots to do anything more than adjusting altitude and waypoints - if things get really ugly there are presently too many variables involved. Sure, AI may eventually be able to handle this but sourcing sufficiently detailed test data is going to be somewhat of a challenge as we tend to avoid problem situations, for very good reasons.

[SloppyJesse]

Hudson, we have a problem

Automated guidance for taxiing around a major airport I get.

But would it determine the Hudson a viable alternative or just plough into buildings attempting to reach the nearest runway?

[ChoHag]

Re: Hudson, we have a problem

Not remotely useful.

If the pilots are all incapacitated, something which basically doesn't happen unless you have a camera crew in the cockpit or are flying alone, the plane could land itself and if the simple act of taxiing is too much for the computer (it isn't) they can send a fleshbag over to do it.

Once it's on the ground, one way or another all the hard problems are solved.

[Jimmy2Cows]

Re: Hudson, we have a problem

Yeah but that wasn't the question.

In essence, if the meatbags are incapacitated and the computer determiines it's not possible to reach the nearest suitable runway, what does it do?

[ChrisC]

Re: Hudson, we have a problem

Assuming there's nothing wrong with the aircraft and it's only the pilots who've been incapacitated - which is fairly fundamental to this discussion given that this is exactly the sort of scenario that Dragonfly is being developed to assist with - then the window of opportunity for incapacitation of the entire flight crew to occur at a time when the aircraft was physically incapable of reaching any suitable runway is tiny, absolutely microscopic, in comparison to the window of opportunity for incapacitation of the entire flight crew to occur at any one of the myriad of other times during the flight when the aircraft WOULD be able to land safely if only it had some way of doing so itself in the absence of any inputs from the flight deck...

So yes, it is possible to concoct some nightmare scenario from which Dragonfly wouldn't be able to save the day, but it's possible to concoct far more, and far more likely, scenarios in which it very much would.

[M.V. Lipvig]

Re: Hudson, we have a problem

If they're smart, and that remains to be seen, the Drangonfly system won't be fully automomous, and if it doesn't see a way out it should ask for guidance from whatever ATC it notified an emergency to. That ATC should be able to provide guidance to the Dragonfly system, and the Dragonfly should be able to carry out those commands even if it doesn't see a solution. An ATC might direct the plane to a water landing that it doesn't want to do.

Now, that being said, the telemetry had better be encoded and hackproof, because the next generation hijacker will hack into Dragonfly and declare the pilots are incapacitated, then will direct the plane to its intended destination or target. It also needs to be built so the pilots can disable it with a physical switch, not software, without disabling the plane's controls. Needs to be a big red light that comes on making it very plain that Dragonfly was activated remotely.

[ChrisC]

Re: Hudson, we have a problem

Apples and Orangutans... Dragonfly is being proposed as a backup in the event of *pilot* incapacitation, whilst the Hudson was an example of *aircraft* incapacitation, where reaching an actual runway via a sensible emergency flightpath wasn't an option. For Dragonfly to need to participate in the latter type scenario you'd already be in the former type scenario, at which point you really are in a spot of bother...

[A.P. Veening]

Re: Hudson, we have a problem

If some of those geese had crashed through the cockpit windows and incapacitated the pilots, that situation would have occurred. Luckily the plane wasn't flying fast enough yet for that to be possible.

[ChrisC]

Re: Hudson, we have a problem

And if that had occurred that day in NY, then it's pretty much a certainty the outcome would have been VERY different not only for all onboard, but likely for some on the ground as well. In such a nightmare scenario, the presence of Dragonfly/A.N.Other automated system isn't going to make matters any worse than it already is, and it would at least offer a slim chance of a more positive outcome if it remained operational following the impact.

The same sort of weird "oh well, if it can't handle extreme scenario X then what's the point of having it at all" argument could be made for pretty much any type of safety system - e.g. airbags in cars. If someone drives their car off a cliff into the raging ocean, there's little or no chance of the airbags being useful in that scenario, but using that as justification for not fitting them to cars would be rather odd given the myriad of other more likely scenarios where they do give vehicle occupants a chance of walking away. Dragonfly is being aimed at giving pilots assistance in some aspects of flight, it's not being promoted (despite the slightly misleading way the article title is worded...) as fully automating every aspect of getting an airliner safely from A to B regardless of what the gods of mischief and mayhem might decide to throw at it along the way.

[An_Old_Dog]

Re: Hudson, we have a problem

The same sort of weird "oh well, if it can't handle extreme scenario X then what's the point of having it at all" argument could be made for pretty much any type of safety system...

My concern isn't that they are adding a safety system: it's that they will use the installation of that safety system as an excuse to remove a safety system -- a pilot-trained human.

[Potemkine!]

If the captain had the fish, for example.

There's also the case of pilots fighting each others in the cockpit.

[Graham Dawson]

The FAA thinks pilots are spending too much time relying on automation, reducing their manual skill and familiarity with their aircraft and increasing the possibility of unchallenged controlled flight into terrain because they trusted the automation. The EASA has taken the reverse position, calling for more automation and less manual control, in order to reduce the possibility of pilot error. Interesting clash up ahead.

[imanidiot]

As a hobby pilot (gliders) I think the FAA is closer to the truth than EASA unfortunately. EASA seems to follow the Airbus line of thought that pilots should only ever operate within the limits of the flight envelope limits set by the flight control system. Accident rates and accident investigations for both Boeing and Airbus aircraft shows that both systems have their drawbacks. Boeing aircraft have done things that they shouldn't have because the pilots were obvious doing something stupid that the computer should have prevented, Airbus aircraft have done things they shouldn't because the computer decided the pilot input was invalid when the pilots wanted it to do something (mostly over confusion of how Airbus autopilot and flight control law modes work and how the system can go from one state to the other)

[A.P. Veening]

Frankly, I am with FAA on this one, especially Asian carriers are susceptible to that risk, see the Asiana crash in San Francisco.

[Lars]

@Graham Dawson

I am quite convinced you exaggerate that difference a lot.

Perhaps you could provide some good links.

And let's not forget that when it's about Boeing and Airbus the opinions tend to be rather bound to geography.

[Graham Dawson]

I've got an article that conveniently contains links to both proposals right here.

I should have linked it before. It's friday. :)

[Lars]

@Graham Dawson

Just silly, being for something is not the same as being against something else.

You know that very well.

[Anonymous Coward]

Airbus pitches the system as an extra layer of safety

to increase airlines' profitability

...

Now, if only there was a way to make passangers bring their own fuel... and seats, and mandatory (bring two, get 5 p off your next flight). And, I dunno, set up crowdfunding before each flight to organize an engine or two? Sharing is caring!

[Michael Hoffmann]

This system will work perfectly...

... until that one day, at JFK, where you need a grand wizard like Kennedy Steve to unravel the kitten-mauled ball of yarn. And the "machine" will steer all planes straight into each other. On the active runway.

[Antron Argaiv]

Re: This system will work perfectly...

I wonder what happens when it takes a Canada goose to the cameras (and both engines)?

Oops...there goes your vision. What are you going to do now?

[ChrisC]

Re: This system will work perfectly...

Rely on the skill and experience of the pilots who'd still be onboard, just as you'd do today in that same scenario...

[Jim Mitchell]

Re: This system will work perfectly...

If you have lost the engines, the cameras and the pilots, then the plane is doomed with out without this new system. It is a silly question.

[Aladdin Sane]

Re: This system will work perfectly...

You got a problem with Canada gooses then you got a problem with me and I suggest you let that one marinate.

[A.P. Veening]

Re: This system will work perfectly...

I wonder what happens when it takes a Canada goose to the cameras (and both engines)?

Oops...there goes your vision. What are you going to do now?

To make it complete, some of these geese will also have to crash through the cockpit windows and incapacitate both pilots. If that doesn't happen, just follow established protocol for landing on the Hudson.

[Graham Dawson]

Re: This system will work perfectly...

God bless that man.

[Anonymous Coward]

remote operation?

I mean, sure, a predator and such birds are 'not exactly' a passanger jet, but in the long term...

(hey, anybody free to take over that 767, it's going down... anybody? Oh I don't know, I guess Dave could give it a go, check the canteen... Anyway, gotta go, see you Monday!)

[Jimmy2Cows]

Re: remote operation?

That does open the risk to the plane being flown remotely by someone less... friendly... than ATC.

[Kildare]

The Other side of this.

Back in the late 80's I was flying to Gatwick from Edinburgh and had the opportunity to visit the flight deck and talk to the captain (different times!)

It was a foggy evening and he told me that the aircraft we were on (BAE 146) was equipped with "Full Auto Land" which could be used to safely land the aircraft no matter how poor the visibility - however due to the cost of testing required to ensure that the system was working correctly it could not be used. The airline had calculated that time lost due to bad weather was less than the recertification.

In the case of this proposal I wonder how the cost of maintaining the system versus cost of a pilot compare - always assuming that they can find people willing to fly with only one pilot.

[First Light]

Not in a strong wind

As someone who has flown (as a passenger) frequently in blustery Irish weather conditions, there is NO WAY I would want the plane autolanded. And blustery weather is the norm over most of the country. I want two human pilots up front and in charge, thank you.

[Duncan Macdonald]

Re: Not in a strong wind

Your plane may well have been landed by Autoland - the pilot would be able to override if needed but automatic landings are now so common that no one notices them. (If the weather conditions are too bad for an automatic landing then the pilots are probably thinking of a go around or even a diversion.)

[Phones Sheridan]

Re: Not in a strong wind

1997 I was holidaying in Tenerife. When about to start his descent the captain started his "we'll be landing shortly" monologue that ended with "we have some bad weather over the airport right now, so we'll let the plane take the strain and perform an auto landing". The panic rippled from the front to the rear of the plane like a mexican wave. People were clearly and vocally uneasy about this, but it's the most calm landing I've ever experienced. We didn't even know we'd landed until the brakes came on. I was told later by an aviation buddy it's because the autopilot can make changes on the controls much quicker than a human, and at a much greater frequency, meaning it compensates before the pilots sitting shotgun are even aware of it.

You have probably experienced an auto landing many times, because it's almost a commodity feature nowadays, and, because of the panic that used to occur, they no longer announce it.

[Anonymous Coward]

Pointless technology for a 737 Max then

Can';t image that it's gonna work for non-FBW types, and then where the Max is concerned it won't need to taxi in a crater.

[steelpillow]

Sanity check

Autoland systems have been available on commercial airliners, from Airbus and others, for a long time. But they rely on suitable equipment being installed at the airport.

This system is just an incremental advance on those, offering less reliance on ground-based systems and hence the opportunity for greater operational flexibility.

It is very much NOT a substitute for a pilot, nor does it reach into AI territory. Airbus are not stupid.

Nor is it as yet production-ready, this is just the first flight prototype being put through its paces.

[ChrisC]

Re: Sanity check

Quite, and reading through several of the other comments so far it feels like this is a rather pertinent point that's being missed/ignored.

It's also worth remembering that Airbus have had a long history of inserting automation/computing elements inbetween the pilot and the control surfaces thanks to their early adoption of fly by wire, so their design ethos is already grounded in such ideas - as you say, Airbus aren't stupid, and between them or Boeing I know which one of the two I'd consider most likely to make a system like this work...

[s. pam]

What would Peter Neumann say?

For those unawares, Peter ran the RISKS Digest for eons in early Interweb days, and before it on email.

The benefit of the use case and expected outcome is clear however what isn't is the journey to safely get there.

[druck]

Re: What would Peter Neumann say?

Some of the most thought provoking articles I've ever read on were on RISKS Digest, and should be compulsory reading for anyone working on safety critical systems.

[Alpc]

What I'd like to know is...

...what would happen if the 'puter was incapacitated, or just busy?

I guess 100% automated planes are still a while off.

[Duncan Macdonald]

Re: What I'd like to know is...

100% automated planes have been around for many years - just not used for passengers. (The first was the German V1 in WW2)

Icon for the end of a V1 flight ======>

[A.P. Veening]

Re: What I'd like to know is...

But the V1 wasn't known for its good landings, the pilot never walked away from it.

[Alpc]

Re: What I'd like to know is...

The V1 never carried many passengers, either.

[Anonymous Coward]

Re: What I'd like to know is...

You thinking of the V3, where the pilot had a parachute and got told that when he had it had on target, to pull the Big Red Lever and he'd be ejected to safety. The teacher was a Judas goat that told them he was an Ace for doing 5 V3s on London.

The Big Red Lever did not eject the pilot. It locked the controls in place, and told the pilot his military life insurance plan was cancelled. The last thing the pilot got was a personal thanks on tape for his sacrifice to zer Fazerlund from old Nutstache.

[midgepad]

V once IIRC

Iirc, a test pilot took a flight on a V1, and did not die, but I've not checked.

"Never"

[RobThBay]

DragonFly vs DraganFly

DraganFly is a long established Canadian drone company.

Draganfly.com

[Kev99]

Once on the ground, it may be a good idea. In the air? Not so much. I flew with my brother once in his Bonanza. Would this system be able to take evasive action as whn some idiot kept flying straight at us

even after my brother flashed his landing lights several times and called out on the radio? Would the system work in areas with multiple bodies of water under the flight path as occurs in Florida, Minnesota and other areas? What about the turbulence caused by one plane ahead of another on take-off and landing? If this were such a great idea, I'm sure the air forces of the world would be falling over themselves to get it.

[midgepad]

Better ask if it can land an F18

On a carrier.

[steviebuk]

..

"I can't let you land Dave".

[AbortRetryFail]

The trouble is...

Just like with cars, the trouble is that the more you take everyday control away from the human, the less capable the human becomes in taking over when the automated systems can't cope.

Granted, with pilots then they can put in the hours on the simulator to keep sharp on handling an emergency, but even so. Plus, you just know that some airlines are bound to scrimp on simulator time.

[LoaderTosh]

Bollards

Wow, that’s amazing … ahh … hang about. No, cancel that, not amazing after all. The Cirrus Vision SF50 single engine personal jet “has that, does that”. In trouble? …, medical or unable to cope with the present situation and conditions? Just push the “I give up” button and the Cirrus will review nearest suitable airports, the suitability of their weather at the expected time of arrival, take you to the most suitable and auto-land you there. There are no “smart electronics for BIG aircraft” … just “smart electronics for aircraft”. And if “that” goes “pear shaped“, the Cirrus Vision SF50 (and their smaller piston engined aircraft) have an aircraft parachute system which will lower you, your passengers and your not inexpensive personal aircraft, gently to the ground. Try that with an A380.

[Anonymous Coward]

Fly Tesla Air

Premium prices.

No staff means we can operate at less of a loss.

[M.V. Lipvig]

Biomimicry?

Don't know about that. Dragonflies jerk in different directions, go up and down like helicopters, don't know if I'd want to be in a plane that's thinking it's a dragonfly even in an emergency.

[Anonymous South African Coward]

"I'm sorry Dave, I'm afraid I cannot land this aeroplane".

[Anonymous South African Coward]

Just one question. American Airlines Flight 191.

Could an AI pilot have saved this plane?

https://en.wikipedia.org/wiki/American_Airlines_Flight_191