Airline Autonomous flight technology

[Clip4]

Which is why we put in our contract that there will be at least 2 pilots in the cockpit

First, you don’t have a contract, you have a collective bargaining agreement, which expires every few years and gets renegotiated.

In 1980, the retail clerks union negotiated no bar code scanners in the collective bargaining agreements and successfully lobbied for state legislatures to profit bar code scanners. Long term, how did that work out for them.

The UAW had similar success with outsourcing provisions in their agreements and tariffs on imports. We all know how that worked out.


I believe many do not understand the world labor shortage that is coming due to populations in the majority of the developed countries not having enough children and the US pulling out of globalization. You are just getting a sample with the supply disruptions today.

 

[Bell206]

You need to get past all the pilots unions, the ATC unions (you’ll not have automated planes without automated atc), public fear of the what if’s, and cost and time to refit.

Those are mainly secondary hurdles to get past. The primary is regulatory and based on defined probability. This starts at the ICAO level and is included in every CAA aircraft certification regulation. But the issue is not what the automation can do, but its failure rate when performing those tasks. So once you can quantify the single pilot failure rate will meet the established probability limits then you will need to deal with those hurdles you mention. I can't post links right now, but last I checked the upper failure limit is 1 x 10 -9. However, ol' Murphy has an historical way of interjecting himself on occasion when new tech is developed and brought to market.

but the FO is going the way of the flight engineer on domestic flights sooner than you imagine.

Sooner is very subjective. The EU and EASA are at the fore front of this move to single pilot ops. But,their form of governing on both sides allows them to move a bit easier on things like this. However, it doesn't move that quick. For example, the EASA finally allowed all small aircraft owners (<6000lbs) to fully manage the airworthiness of their aircraft in 2020 which is something owners here have had since 1920s. Regardless, I think once you hear of or see an STC to SPIFR a common pax airliner model or a new aircraft designed for single pilot then this discussion can move forward from what I see being discussed now.

the US pulling out of globalization.

Do you have a reference for this? Its not what I see/read/hear of. If anything they are expanding its reach by spreading global requirements over more countries. At least on the aviation side.

 

[Salty]

Pretty sure that in the early 1990s I was using an Amiga computer with a 300 (ok maybe 600) baud modem to dial into other peoples computers over the phone with BBS’s software on them. That was the state of the internet for most people back then. Compare that to today where things like Google and chat gtp can be accessed from any cheap $50 phone that can be carried in your pocket. All that happened in less than 30 years. Nobody in the early 1990s saw the internet turning into what it is today. How can you be sure airline automation won’t also advance and become common in the next 30 years?

I think you're about a decade off. Most of us were dialing into ISP's with 2400 baud+ by 1993.

 

[tspear]

Sure, they need those pilots now, but sitting in airline Board rooms are the people who will evaluate all the ways to reduce that work force after they have hired them and lobby Congress to make regulatory changes to allow this to happen.

I doubt in my life time I will see pilotless aircraft, but the FO is going the way of the flight engineer on domestic flights sooner than you imagine.

One issue which has not been addressed, is the suicide rates. Until that issue is addressed, you will likely see significant resistance to single pilot operations.
If you look at the financial reports of most of the airlines, airplane crew is piece of the cost puzzle. Capital requirements, fuel, MX... all of these are significantly higher.

Tim

 

[jordane93]

First, you don’t have a contract, you have a collective bargaining agreement, which expires every few years and gets renegotiated.

In 1980, the retail clerks union negotiated no bar code scanners in the collective bargaining agreements and successfully lobbied for state legislatures to profit bar code scanners. Long term, how did that work out for them.

The UAW had similar success with outsourcing provisions in their agreements and tariffs on imports. We all know how that worked out.


I believe many do not understand the world labor shortage that is coming due to populations in the majority of the developed countries not having enough children and the US pulling out of globalization. You are just getting a sample with the supply disruptions today.

I see it coming eventually just not as quick as you may think. Also you can’t compare retail workers to pilots. It’s much easier to automate a cashier’s job. I think people would be skeptical to see one pilot up front.

 

[Groundpounder]

I have 21 years of my airline career left. I don't think I will have any issues. But if had 40 years left, I would be concerned. They will first come up with a way to reduce augmented crews, then airplanes will go single pilot, then pilotless. Thankfully the unions are cognizant of such efforts and are putting up the good fight, but I think it may be inevitable. Of course one accident early on could change everything.

 

[kmacht]

30 years ago you couldn’t go to your mailbox without finding an annoying AOL cd in it. And, while crude to todays standards, you weren’t “connecting one at a time to some guys basement”.

30 years ago was 1992. The internet itself existed in the 80s but only in academic and government settings. The World Wide Web wasn’t invented until 1989, the first browser (mosaic) wasn’t released to the public until 1993 and AOL didn’t start sending out CDs until 1994which were little more than some online games and access to news stories. 1990 and 1991 were still very much the years of dial up into people BBS’s on their computers at home. There were some that were dabbling in faster speeds and platforms like compuserve but most of the countries experience with the internet n the very early 90s was still with slow dial up modems directly into other people computers.

All of that has nothing to do with the point I was originally trying to make though. The internet has changed drastically over the past few decades regardless of the exact time periods being debated. To think that automation and flying isn’t going to or can’t advance to the point of general acceptance over the next 30 years is just not logical.

 

[Larry in TN]

First, you don’t have a contract, you have a collective bargaining agreement, which expires every few years and gets renegotiated.

Airline labor relations are regulated differently than most other industries. They are covered by the Railway Labor Act (RLA). Under the RLA, labor agreements never expire, they become amendable. They remain fully enforceable indefinitely, unless a new agreement is ratified.

That said, I agree that a two-pilot CBA provision will not hold up for long. It was tried back when newer airplanes were being certificated without a flight engineer. Some airlines did fly around with flight engineers in two-pilot airplanes for a while, but that ended fairly quickly.

Many who argue that single-pilot, and pilotless, airliners are closer, rather than farther, away is that they tend to focus on the physical piloting of the aircraft and not on all the other tasks that human pilots perform. Physically flying the aircraft is the easy part to automate. Single-pilot is particularly problematic because you haven't automated all human tasks, you would be pilotless if you had, but you've removed the backup, CRM, threat-error management, VVM, workload sharing, etc. that comes from crew interaction. That is the part that will be difficult to automate.

 

[NorthEast Ohio]

In the 1950s and 60s NASA spectacularly blew up dozens of rockets working out the bugs, racing the Soviets to the moon.
My question is…
Who, today, is going to risk spectacularly crashing a few airliners before they work out the bugs of autonomous aviation?

 

[Pi1otguy]

It would be rare for the autopilot to be engaged that early, at least on US airlines

I like to believe that the big iron always gets "as filled" from clearance delivery. And that the magically FAA servers have designed a slot in time and space for them such that the radio chatter with tower and departure are mere formalities.

 

[Groundpounder]

I like to believe that the big iron always gets "as filled" from clearance delivery. And that the magically FAA servers have designed a slot in time and space for them such that the radio chatter with tower and departure are mere formalities.

Definitely not! Especially in complex airspace, and doubly so when there is weather.

 

[Larry in TN]

I like to believe that the big iron always gets "as filled" from clearance delivery. And that the magically FAA servers have designed a slot in time and space for them such that the radio chatter with tower and departure are mere formalities.

My favorite is when you are given an unsolicited shortcut, a frequency change then a speed reduction, another frequency change then delay vectors, another frequency change then a shortcut and "max forward speed".

 

[asicer]

Pretty sure that in the early 1990s I was using an Amiga computer with a 300 (ok maybe 600) baud modem to dial into other peoples computers over the phone with BBS’s software on them. That was the state of the internet for most people back then.

First off, that's not "the internet". Second, I think you are off be a several decades.

I could go on, but...

 

[kaiser]

You all remember trust falls at camp? That's what I think about when I see "autonomous" flight (or cars or whatever).

I hated trust falls.

 

[Salty]

That’s a misnomer. I call them “sucker falls”

 

[dmspilot]

Aren't most flights on autopilot above 1000' AGL as is? I figure the pilot is there to question ATC (see FedEx vs Southwest at Austin) and as manual reversion when gps and the LOC are down.

The autopilot might be wiggling the stick, but the pilots are "flying" the autopilot. A departure or arrival, especially one where ATC can't make up their minds whether they want us to speed up or slow down (and if neither of those work, start issuing panic vectors), requires near constant and continuous intervention.

 

[Clip4]

Airline labor relations are regulated differently than most other industries. They are covered by the Railway Labor Act (RLA). Under the RLA, labor agreements never expire, they become amendable. They remain fully enforceable indefinitely, unless a new agreement is ratified.

That said, I agree that a two-pilot CBA provision will not hold up for long. It was tried back when newer airplanes were being certificated without a flight engineer. Some airlines did fly around with flight engineers in two-pilot airplanes for a while, but that ended fairly quickly.

Many who argue that single-pilot, and pilotless, airliners are closer, rather than farther, away is that they tend to focus on the physical piloting of the aircraft and not on all the other tasks that human pilots perform. Physically flying the aircraft is the easy part to automate. Single-pilot is particularly problematic because you haven't automated all human tasks, you would be pilotless if you had, but you've removed the backup, CRM, threat-error management, VVM, workload sharing, etc. that comes from crew interaction. That is the part that will be difficult to automate.

Another issue is since deregulation non union airlines would have a certain advantage over the union carriers bound by a two pilot crew. Bankruptcy legally sweeps away CBAs.

 

[Groundpounder]

Another issue is since deregulation non union airlines would have a certain advantage over the union carriers bound by a two pilot crew. Bankruptcy legally sweeps away CBAs.

I think the only non union carrier with any significant fleet size is Skywest, and they’re “only” a regional.

 

[Clip4]

I think the only non union carrier with any significant fleet size is Skywest, and they’re “only” a regional.

That’s what the UAW workers said about Toyota.

 

[Groundpounder]

That’s what the UAW workers said about Toyota.

They said that Toyota is a regional carrier that depends on a larger carriers feed, and doesn't have any of their own, independent flying? That is an odd thing for people that build cars to say, but OK.

 

[Clip4]

They said that Toyota is a regional carrier that depends on a larger carriers feed, and doesn't have any of their own, independent flying? That is an odd thing for people that build cars to say, but OK.

No they said they were a small non UAW operation.

 

[OneCharlieTango]

My favorite is when you are given an unsolicited shortcut, a frequency change then a speed reduction, another frequency change then delay vectors, another frequency change then a shortcut and "max forward speed".

Good point. It seems to say that AI for ATC is a necessary prerequisite for AI in the cockpit. Interface the two, and there’s no human in between to frustrate.

 

[Shuckers86]

Everyone on that SWA 737 in KAUS should be very thankful there were real pilots in that FedEx 767. The pilots, not the airplane, chose to discontinue the approach in a low vis situation. Using their eyes, training, and judgment they avoided a catastrophe. A drone 767 might have continued to a landing like it was cleared and programmed to do. Nicely done FedEx crew! Bravo Zulu!

 

[tspear]

Everyone on that SWA 737 in KAUS should be very thankful there were real pilots in that FedEx 767. The pilots, not the airplane, chose to discontinue the approach in a low vis situation. Using their eyes, training, and judgment they avoided a catastrophe. A drone 767 might have continued to a landing like it was cleared and programmed to do. Nicely done FedEx crew! Bravo Zulu!

Why do you make the assumption the unmanned system would have continued the approach?

Tim

 

[jsstevens]

From psychological perspective, AI for planes (and more importantly for cars) will not have to be just better that humans. We have an irrational fear of a machine killing us. (It's not irrational to fear being killed, just that we trust humans more than we trust machines even if the machines are statistically safer.) Notice how far and wide corner cases where the humans saved the situation get broadcasted vs. the vast majority of cases. Humans are motivated by stories much more than by data. It takes a certain mindset and mental discipline to look at data and change your behaviors accordingly.

 

[Larry in TN]

Using their eyes, training, and judgment they avoided a catastrophe.

I'd rate judgement and situational awareness as the most important factors.

It is clear that the FDX crew saw the situation developing early based on them confirming the landing clearance. They did that because they doubted that there'd be enough room for the departure. Neither the controller nor the SWA crew gave any indication, at that point, that they understood the developing situation.

Why do you make the assumption the unmanned system would have continued the approach?

How would the AI have known that it shouldn't land?

This is what I addressed up-thread. Automating the physical handling of the airplane is the easy part. A 767 does a great job of landing itself when everything is setup correctly. Situational awareness, threat management, error identification and repair, etc. is the hard part.

 

[Half Fast]

This is what I addressed up-thread. Automating the physical handling of the airplane is the easy part. A 767 does a great job of landing itself when everything is setup correctly. Situational awareness, threat management, error identification and repair, etc. is the hard part.

Spot on.

I don't think it's possible, at least not in the foreseeable future, to have an extensive enough data set to train the automation to have sufficient judgement to handle all the corner cases that only pop up once in a pilot's lifetime.

The Miracle on the Hudson was possible because both Sully and Skiles had thousands and thousands of hours flying just about anything with wings. They also had many experiences outside of aviation that had developed their senses of judgement. Human perception and judgement begins in the cradle, not when a person first sits in a cockpit. You simply cannot train automation to think the same way and make similar good decisions in all the extreme cases.

Flying is the easy part. I've helped design cruise missiles that can fly tremendous distances and deliver a warhead through a particular window in a particular building. But there's no way on earth I can give one the judgement to assess world events and decide when to launch itself and how to pick out its own target.

 

[Bob Noel]

to piggy back...

imagine the challenge of automation handling the UA flight 232 failure and subsequent "landing" (July 1989)

Edit: And by putting quotes around the word landing, I meant no disrepect to the incredible job they all did getting the airplane down.

 

[Larry in TN]

imagine the challenge of automation handling the UA flight 232 failure and subsequent "landing" (July 1989)

For those who don't know the flight number, that's the DC10 that lost off hydraulics, which means they lost all flight controls, as well as one engine. The crew, plus a deadheading pilot/instructor, managed to land it without flight controls and with over 60% of the occupants surviving.

 

[tspear]

@Half Fast

Are you sure the thousands upon thousands of hours sitting in cruise helped?
I would think it was the training, life experience, their fight/flight inclination, and any flying not involved with airlines.
Note: I am not an airline pilot, but from everything I have read, airlines never train/plan for a complete engine failure.

Tim

 

[Greg Bockelman]

For those who don't know the flight number, that's the DC10 that lost off hydraulics, which means they lost all flight controls, as well as one engine. The crew, plus a deadheading pilot/instructor, managed to land it without flight controls and with over 60% of the occupants surviving.

Denny Fitch, anyone?

 

[Larry in TN]

Denny Fitch, anyone?

Yep. I never got to meet him. Saw the extended interview that he did about the flight.

Here's a great book that covers the accident in depth. It details the history of the titanium slug that because the disc that failed, how the crew, cockpit and cabin, performed their duties, and personal information about the occupants.

www.amazon.com/gp/product/B00FQUDSFS

 

[Half Fast]

Are you sure the thousands upon thousands of hours sitting in cruise helped?

No, I think it was all the hours spent in gliders, taildraggers, seaplanes, etc., plus all the hours giving flight instruction, that helped. Both those guys lived and breathed flying and spent lots of time flying other craft and participating in non-flying aviation activities, all of which honed their minds and problem-solving skills.

But more important were all the non-cockpit hours that influenced their ability to assess a situation and make decisions, hours they accumulated all through their lives. I think you skipped this part of my post: "They also had many experiences outside of aviation that had developed their senses of judgement. Human perception and judgement begins in the cradle, not when a person first sits in a cockpit."

You cannot train that sort of judgement into an automated system.

 

[Greg Bockelman]

Yep. I never got to meet him. Saw the extended interview that he did about the flight.

Here's a great book that covers the accident in depth. It details the history of the titanium slug that because the disc that failed, how the crew, cockpit and cabin, performed their duties, and personal information about the occupants.

www.amazon.com/gp/product/B00FQUDSFS

I had the pleasure of flying with him twice before he retired. Very down to earth guy. He does not get the credit he deserves for his contribution to that flight.

 

[sourdough44]

Yeah, I can Handel 320 knots through moderate turbulence. I guess if the flight plan calls for it, go for it.

 

[Palmpilot]

Yeah, I can Handel 320 knots through moderate turbulence. I guess if the flight plan calls for it, go for it.

I prefer to Mozart it.

 

[Groundpounder]

No they said they were a small non UAW operation.

apples/oranges.

 

[kmacht]

For all the examples of a pilot saving the plane in extremely odd and unlikely situations how many are there of pilots screwing up and crashing. The NTSB database is loaded with pilot error listed as the cause.

 

[Bell206]

For all the examples of a pilot saving the plane in extremely odd and unlikely situations how many are there of pilots screwing up and crashing. The NTSB database is loaded with pilot error listed as the cause.

True. But any comparison needs to be kept in context: human error vs computer error. And I believe if you were to dig a little deeper in the data you'll find computer system failures far exceed any direct human error caused accidents.

 

[Palmpilot]

True. But any comparison needs to be kept in context: human error vs computer error. And I believe if you were to dig a little deeper in the data you'll find computer system failures far exceed any direct human error caused accidents.

I'm not sure that's true, but I'm not sure it's relevant, because the range of tasks they're called on to perform is quite different. If we're going to compare the rate of human error to the rate of computer error, we need to make the comparison with both of them performing the same tasks, in this case, performing the duties of the pilot-in-command of an airliner.