Automation - State of the Art

[Dont Hang Up]

There is much talk of the high degree of automation in modern aircraft. I am interested to understand what point this has reached.

If we could ignore ATC and other traffic separation constraints, what are the minimum possible flight crew interventions required to carry out a typical flight from the line-up for take-off to the landing roll-out?

[John Farley]

Don’t Hang Up

I think it might help in answering your question if we separate the two jobs crews perform:

1 Steering the aircraft (controlling speed, altitude and heading)
2 Operating it (every other function).

For my money the current technology can/could cope with all steering since it is already certificated for automatic landing – arguably the hardest and most demanding steering function.

As to operating the aircraft (not just dealing with ATC) the crew will have to intervene throughout the flight in many different ways depending on the type of flight, the type of aircraft, the type of company, the type of passengers etc etc. In my view there is no current technology that can carry out these functions without considerable crew intervention nor is there likely to be in the foreseeable future.

Indeed I would go further and say we should not waste money on trying to automate these other functions (other than ATC aspects) because humans are so good and flexible in carrying out such duties. Plus we will always need people locked in the front or nobody would (sensibly) ever get in the back.

[muduckace]

Technically viable,

1-FMS programmed via WIFI at the gate, aircraft serviced.
2-Engine start could be automated or performed by ground personell
3-Bush back with ground commands for brakes - taxi with GPS guidance and TCAS, proximity sensors as backup or supertug to the threshold.
4-Automate flaps, gear, spoilers etc.. (not hard)
5-FANS allready exists next step would be commands instead of messages, FMS could command T/O rollout by FANS command interfaced with FMC's etc.
6-FANS could continue to command an aircraft, flight level changes, wxr avoidance to tech failure resolution procedures, passing off the aircraft from controller to controller commanding changes.
7-intercept ILS, controller arms land mode and switches to another controller who just monitor approach and rollout ready to hit G/A and pass back off to tower.
8-finally steps 4,3 and 2 back to the gate.

All ground controller and most handeling steps could of cource be automated in the future. I see this coming one day first with 1 pilot monitoring the aircraft performing a series of required attention tests to make sure he does not fall asleep.

I say to all the naysayers, what would pilots have said back in the 60's about removal of the F/E's position, or years before that of the navigator.

[Dont Hang Up]

My understanding of these things has always been of a large number of automated sytems and subsystems, with the flight crew acting as the glue to hold it all together (and taking over completely if any of them fail).

However, based on one or two of the discusions appearing on PPRUNE I had begun to think my understanding was falling behind and maybe the automation was becoming a lot more joined up than I believed. Hence my question.

I could be tempted to express it slightly more flippantly (it is Friday afternoon so forgive me).
If the Captian and FO select take-off power and retire to the cabin to play bridge with the first class passengers, how many rubbers could they realistically get in before catastrophe strikes?

:)

[muduckace]

Perfect world (flaps and gear up too), the answer would be until the aircraft fly's its self into the ground for most modern airliners as it fly's into the programmed STAR and there is no one there to lower the gear and flaps and arm autoland, may just run out of fuel. Depends on type.

Could have a very long card game and get soo ****faced they pass out for the big show.

Perfect world assuming they don't get shot down as well.

[rottenray]

Don't Hang writes:

how many rubbers could they realistically get in before catastrophe strikes?:eek:

This is off-topic and strictly for your education, Hang.

Here in the colonies, "rubber" is the slang term for condom.

I can now dry my office chair, thank you for the chuckle, and get back to work!



Back O/T...

The interesting thing about this recurring discussion is the history of automation itself, across all industries.

It has gone from replacing repetitive manual labor which was hard to hire for and harder to retain trained employees for, to replacing less-accurate human assembly with more accurate robotic assembly, to replacing skilled humans and semi-skilled for the sake of cost savings.

What differs in the aviation world is that the "growing pains" of automation in other industries is generally less disaster-prone - a robot on a factory floor can mess up, and usually all that is lost is a bit of product.

Certainly, the technology is already available to automate flight.

Doing so represents a very complex system.

Referring to Mud's post (http://www.pprune.org/tech-log/414519-automation-state-art.html#post5679818), I think we can consider possible comms failures and how to cope with them as the major stumbling block.

Everything in the air, and on the ground waiting to take off, is subject to the whims of a plethora of variables - block length changes due to winds at cruise, et cetera.

"De-bugging" this system will be fascinating to watch.


RR

[411A]

Modern day aircraft automation with large jet transport aircraft started with a fine design from Lockheed, and progressed in a measured reasonable way, with later Boeing types.

Pilots...'go away?'

Hardly likely.

[joe two]

If the Captian and FO select take-off power and retire to the cabin to play bridge with the first class passengers, how many rubbers could they realistically get in before catastrophe strikes?



None,

as the remainings of that aircraft would lie burning , somewhere in the extension of the runway.

[GlueBall]

The USAF won't disclose the embarrassing number of multimillion dollar UAVs [Unmanned Aerial Vehicles] that have failed to returned to base. . . !

As to automated passenger flights: all is well as long as there are no system failures, nor other phenomena, such as moderate clear air turbulence which could kick off the autopilot(s) . . . .

Are you ready to get aboard an airplane without a pilot...? :confused:

[Dont Hang Up]

Are you ready to get aboard an airplane without a pilot...? http://images.ibsrv.net/ibsrv/res/src:www.pprune.org/get/images/smilies/confused.gif


Definitely not. But that wasn't really the point of my OP. It did not want to restart the debate on pilotless aircraft here on the Tech Log as that is being done elsewhere. I just wanted to understand better the current capability of automation.

Opinions seems to differ.

[SFI145]

411A said
Modern day aircraft automation with large jet transport aircraft started with a fine design from Lockheed, and progressed in a measured reasonable way, with later Boeing types.
In fact the first autoland flight with passengers was carried out by a BEA De Havilland Trident 1 in about 1964 I think

[rottenray]

411a writes:

Pilots...'go away?'You are most likely right. Or at least right for the foreseeable future.

But it all leaves the question of skill, which is developed through practice.

A fair number of the professional pilots posting here on PPRune have commented that handflying time is already harder to get than it was even a decade ago.

Here's an honest question: Will the improvement and growth of automation result in better sims with higher fidelity, or are we at the [reasonable] peak currently?

RR

[Jetex_Jim]

Are you ready to get aboard an airplane without a pilot...? http://images.ibsrv.net/ibsrv/res/src:www.pprune.org/get/images/smilies/confused.gif

Not bl**dy likely.

Robert Heinlein, the great SF and Fantasy writer, predicted that even in the far future human pilots would be maintained. The way he put it, as long as you have a human pilot on board, no matter how bad the catastrophy, and when all the logical possibilities have been exhausted, the human pilot will keep trying. (and perhaps find a solution amongst the non-logical possibilities.) By way of example, Jim Lovell explained how he followed the biolumiscent trail behind his carrier when all the navigation systems had failed.

And that's the way I'd want it to be, thank you very much.

[cosmo kramer]

DHU,

If we could ignore ATC and other traffic separation constraints, what are the minimum possible flight crew interventions required to carry out a typical flight from the line-up for take-off to the landing roll-out?

For the case of 737 the below scenario would be possible. I see from your profile that you are a PPL, so I'll explain a bit of the 737 technical stuff as well, in simple terms.

After manually taxiing to the runway, arm autopilot mode LNAV and VNAV = Lateral/Vertical Navigation where the autopilot is directed by the pre-programmed Flight Management Computer (FMC) to follow a route and a climb, cruise and descend profile. LNAV is basically GPS guidance as your would find in your car. VNAV controls the aircraft path by commanding the speed and throttles depending on the phase of flight and is thereby able to accelerate the aircraft during the climb and vice verse during descend.

The takeoff is always made manually (but with auto throttle engaged for the most cases), but I have heard that it has been tried in the simulator to engage the autopilot, before initiating the takeoff by activating the autothrottle. If indeed true the autopilot could rotate the aircraft by itself. The pilot would still have to steer the aircraft along the runway with the rudder pedals. Once airborne the autopilot would keep wings level and control pitch to keep the selected takeoff speed. As the engines produce takeoff thrust, the autopilot would have to pitch the nose up to keep the speed from accelerating thus resulting in a climb. Passing 400 feet the armed LNAV will engage, and practically direct the aircraft almost to it's destination. VNAV would engage after the FMC pre-programmed acceleration altitude and with flaps up (in my company we do not arm VNAV on ground so not absolutely sure about the criteria - but the point is that it's possible to for this to happen automatic). Hence for the take off only the flaps and gear would have to be retracted manually.

The climb, cruise and descend would then follow the FMC route and profile. Though for longer flights, were a higher cruising level is required as the fuel is burned, there has to be an intervention as well. The descend would also have to be initiated by someone. VNAV would then descend the aircraft and slow it to the minimum speed before flap extension is required.

For auto landing, someone has switch the autopilot from LNAV/VNAV to approach mode for the autopilot to follow the ILS. The complete radio setup for this could be made prior to departure, as this is not required for takeoff (though in real world it would be set as backup for the departure routing in case of LNAV failure and for crosschecking). Only things that would need to be done is arm the autopilot approach mode, arm the autobrakes, select the landing speed in the FMC and engage the second autopilot (both autopilots required for auto landing). And of course someone to extend the landing gear and flaps. When the approach mode engages you also have to set the autopilot speed manually. The autopilot would land and (if chosen as an options from Boeing) do the roll out on the runway - itself reducing the throttles to idle and braking to a complete stop (no reverse thrust though).

As you see very little intervention from crew needed, hypothetically!

If only it worked as advertised! But the fact that the FMC has the same computing power as a Commodore 64 from 1984 doesn't make it so. The biggest factor in the equation being the weather. See, if there is a bit more tailwind than expected during descend for example, VNAV will not be able to cope and will require that the speedbrakes be extended (manually) and in many cases it will not be sufficient and another mode must be used and another profile be flown with the "less advanced" autopilot modes.

I'm drifting away from you question about minimum crew intervention, but nevertheless it's so hypothetically. Even if like you said we pretend that we are the only aircraft in the air.

About the inevitable question about pilotless aircrafts.
In my opinion weather and general decision making are the biggest factors in flying a modern airliner. Yes with very few means, it would be possible to upgrade the automatics to retract the gear and flaps itself and the other smaller tasks. And with some descent programming and modern computing power probably also very easy to make the VNAV handle unexpected tailwind etc. Maybe even make is so reliable that it no longer would be necessary for the pilot to have the skills to handfly anymore (absolutely not so today reliability wise, far from). But who is going to decide whether or not to divert if there is a sick or unruly passenger onboard. Who will make the decision to continue to a remote destination in case of minimum fuel. And so on... Pilotless aircrafts in our lifetime? No chance! When you include traffic which we ignored, then even more unlikely. Even if ATC could pass directions to the pilotless aircraft, I would be very reluctant to board such an aircraft in Europe as a passenger. Now try to trust the same ATC passing over a third world country (will not offend anyone by mentioning any countries, but ATC wise even some countries in Europe are also third world like). Aircraft that will be flow on autopilot 100% of the time maybe, but unlikely as this would require massive investment in ground equipment at airports, and if keeping the same amount of traffic, then possibly development of completely different equipment that we have today (auto land requires that there are no other aircraft near the landing runway, not to interfere with the radio signals). And all this for what purpose? If you are going to pay two guys to sit there anyway to take directions from ATC, check and evaluate the weather along the flight, keeping track of fuel, system operability and performance and so on - then why not let them work during the takeoff and landing and save all that money.

[muduckace]

The way he put it, as long as you have a human pilot on board, no matter how bad the catastrophy, and when all the logical possibilities have been exhausted, the human pilot will keep trying.

This is why I see a transition over a long period of time down to one pilot, then to ground control making human decisions but do not discount eventually the possible evolution of UAV passenger flight.

We would have to see major advancement of environmental sensory and the logic needed as a result to replace human inputs completely but it is possible.

One other thought, this speaks of society in a whole. liability... Every day people make more incorrect decisions as a result of variables in their decision making abilities due to liability.

[engfireleft]

During the course of even the most routine flights a crew will use different levels of automation in different ways to accomplish different goals. Right down to turning it off completely if it isn't doing the job in a dynamic situation. An indication of how well a pilot understands the aircraft is when they know what level is appropriate at any given time, and can anticipate what level will be required in the immediate future. That takes judgement, knowledge and experience.

Some things you just can't program into a computer.

[cosmo kramer]

muduckace,

There is a world outside nice and homogeneous USA. Here in Europe there are 50 different countries in roughly the same geographical area. Different languages, culture, attitude towards safety and regard for human lives. ATC controlling aircrafts will never happen, thus there will always be the need for someone in the aircraft to take (and evaluate and possibly decline) instructions from ATC. If one guy is needed, there has to be a backup in case he becomes incapacitated or dies. Hence, 2 pilot crew will persist for a foreseeable future. I doubt there will be a change to that in our time.

In fact I think this is as likely as world peace.

[Pugilistic Animus]

YouTube - Airplane! (Movie) - Autopilot (http://www.youtube.com/watch?v=vm7i2825dg8)

We all knew serious discussion would not continue indefinitely:}

:ouch:

[muduckace]

There is a world outside nice and homogeneous USA. Here in Europe there are 50 different countries in roughly the same geographical area. Different languages, culture, attitude towards safety and regard for human lives. ATC controlling aircrafts will never happen, thus there will always be the need for someone in the aircraft to take (and evaluate and possibly decline) instructions from ATC

I believe FANS was developed to eventually remedy that nasty little communication problem you have in Europe. It is all over the world, I have experienced it, worst influence is a native tongue in a native land then ATC switching back to english to handle us.

A recent crash that this was probably an issue but not made a big deal of was the EMB v/s 737 over the Amazon. Used to hear altitude and heading reported in Portuguese all the time, you can do a good job of avoiding an RA or worse if you hear another aircraft reporting he is at the same FL heading towards you by questioning ATC.




I can see a single pilot in the next 30 years operating a completely automated aircraft pressing a button every 15 minutes just to tell ATC he is awake. He would do a prescribed amount of T/O's and landings but mostly just be there until the human factor in aviation was trusted as not necessary.

We are testing some aspects of what will be today with military UAV's. I can immagine unmanned freighters first then pax. When.. Who Knows?

I realize what I speak of is more or less science fiction today but look at the history of how dreams soo often become reality.




During the course of even the most routine flights a crew will use different levels of automation in different ways to accomplish different goals. Right down to turning it off completely if it isn't doing the job in a dynamic situation. An indication of how well a pilot understands the aircraft is when they know what level is appropriate at any given time, and can anticipate what level will be required in the immediate future. That takes judgement, knowledge and experience.

Some things you just can't program into a computer.


This is only applicable to the technology and procedures we use today, this will change, just as it has been changing over the last century more or less. Anything can be programmed, you do not fly at your will you do so following teachings and procedures. Teachings and procedures are nothing more than a set of logic tree's "programming".

[411A]

In fact the first autoland flight with passengers was carried out by a BEA De Havilland Trident 1 in about 1964 I think

Could well be, however, aircraft systems automation is not just restricted to automatic approach/land (autoland) ops.

[cosmo kramer]

muduckace
I believe FANS was developed to eventually remedy that nasty little communication problem you have in Europe. It is all over the world, I have experienced it, worst influence is a native tongue in a native land then ATC switching back to english to handle us.

It not just communicating. Cultural differences are more important. Do you want to be a passenger in an aircraft controlled by Romanian ground crew?

Secondly (hence the comment on world peace) some places the neighboring countries are not even acknowledging each others existence. In Europe for example, the is no ATC coordination or handoff between Turkey and Cyprus.

[olster]

The human brain still has a lot more computing power than the most sophisticated computer systems.As much as the bean -counting world would disapprove,commercial aircraft without pilots is a long way ahead.

[RalphTheMouth]

Got this in an email....

From a CAL pilot friend;

"Steve Chealander member of the NTSB 2007-2009 is a retired American Captain. He gave a safety presentation at recurrent training about two years ago. He opened the floor for questions and one guy asked facetiously when are we going to one pilot cockpit?

Chealander said that is not funny. He said Fed Ex and UPS are now, (two years ago), working on the procedures for a one-pilot long haul over-water operation. The pilot would be at the controls for take off and landing then go to the bunk for cruise while the guy back in Memphis would take over for the cruise. One pilot passenger flights will take a bit longer to get approved.

Twelve years ago, I was Director of Operations for the Alaska Air Guard. I went to a high-level conference and this three-star General gave a presentation that said the exact same thing the major said. The only limitation on fighter aircraft now is the pilot. We have the technology to do everything from the ground and it will be a huge cost savings. No search and rescue, no life support systems, no backlash when we lose a plane. So this article is right on the money.

I attended a flight safety presentation last evening from a retired AF Major test pilot from Edwards, Bill Koukourikas, now serving there as a civilian. During the course of his presentation, his statement, "No future attack military aircraft within the next 15 years will have pilots in the cockpit. The last tactical aircraft with a pilot in the cockpit will be the F-35."

He also indicated that, within the next 10 or so years, all UPS and FedX cargo flights will be with pilot-less aircraft. This prediction comes from their test shop at Edwards. All drone testing, development , etc., is taking place just south of Edwards in the Palmdale area. Sounds like a continuation of the Skunk Works developments of Lockheed which previously took place in that area."

Simply amazing! Hey, are we a dying breed or what?

Guess you'll have to invite your computer to "have a beer" after the day's flying is done. Oh, I forgot. They're taking that one away too!

[rudderrudderrat]

Hi,

Strange how we need double dual or triplex autopilots to do CAT IIIB automatic landings, with the equipment cross monitoring one other. Yet some think that a single human needs no cross monitoring.

Computers are only as good as the software writers' imagination. How many of us have had to reset any aircraft's computer?

We'll never fly single crew for commercial passenger operations - there will always be two of us - on half pay though.

[FullWings]

I don't think there is any technical reason why we couldn't one day have pilotless airliners - I feel it is more about cost and being human.

We are complex biomechanical computers that need a considerable amount of programming and environmental exposure to become competent at all respects of operating an air transport. The level of AI required to replicate this I would posit as being "strong". Once this level is reached, we're probably into the scenario of a "technological singularity" anyway, so not much point worrying! Anything a human could do could be done better by an AI...

Being a good pilot has long been about more than the physical aspects of aircraft control; indeed, automation of these functions has been around for a significant part of the history of aviation. The cognitive aspects of decision making have been studied for a much shorter period, however.

[Dont Hang Up]

I attended a flight safety presentation last evening from a retired AF Major test pilot from Edwards, Bill Koukourikas, now serving there as a civilian. During the course of his presentation, his statement, "No future attack military aircraft within the next 15 years will have pilots in the cockpit. The last tactical aircraft with a pilot in the cockpit will be the F-35."

The primary case for UAVs in the military is to remove the pilot from a position of danger thus allowing for higher risk missions, together with lower cost and more dispensable airframes.

Try that argument on your typical fare paying passenger! :eek:

[Dont Hang Up]

Don't Hang writes:

how many rubbers could they realistically get in before catastrophe strikes?


This is off-topic and strictly for your education, Hang.

Here in the colonies, "rubber" is the slang term for condom.

I can now dry my office chair, thank you for the chuckle, and get back to work!


Oops! :O

Intriguingly, a double entendre that works equally for male and lady flight crew.

[Basil]

Automatic approaches to autoland are more demanding than a manual approach. The reason is that the pilots have to understand various complicated failure modes and how to handle them. Sometimes they can be recovered and sometimes they can't. I won't confuse the uninitiated by going into detail; the cognoscenti (Sp?) are too well aware.
As 411A will confirm, the L1011 flies an amazing auto approach, a technique pioneered and perfected by the Brits.
That's all just the final approach; how about the rest of the flight?
Think about en route Cb avoidance, local ATC taking offence and intercepting. Does the on-board computer understand what a fighter pilot means by flying alongside followed by a sharp break? Yes, a ground controller may understand; can we guarantee an uninterrupted datalink?
A 'difficult' country tried to deny us overflight at short notice. Turning back would have been hazardous and the pax wouldn't have been taken to destination. Our understanding of the local psyche and assets followed by an apology permitted a continued flight.
I could go on for paragraphs.

IMHO we are nowhere near an aeroplane with pax & CC flying with no pilot on board.

[peterpuck]

I've never understood the "but if the airplane can land itself, why are you there?" question. To me landing is the easy part. The decisions made prior to push back until landing are the hard parts. It's like saying "the nurse can put in stitches, so why do we need the heart surgeon?".

[aterpster]

GlueBall:
As to automated passenger flights: all is well as long as there are no system failures, nor other phenomena, such as moderate clear air turbulence which could kick off the autopilot(s) . . . .


1. No aircraft on a collision course with an operative transponder.

2. No aircraft on a collision course with an inoperative transponder.

[aterpster]

Strange how we need double dual or triplex autopilots to do CAT IIIB automatic landings, with the equipment cross monitoring one other. Yet some think that a single human needs no cross monitoring.

Not strange at all. We are asking the automatics to gently "collide" with the ultimate obstacle (the runway) when the runway cannot be seen, something we do not expect a pilot to be able to do when hand-flying the aircraft.

[rudderrudderrat]

Hi aterpster,

The automatics can "see" the runway using ILS & Rad ALT information. We gently collide with the runway when we do a visual landing.

I think the two are very similar - so that's why I think we need two pilots.

[Denti]

Well, so i arrived at the aircraft today and the maintenance staff told me everything is ok, absolutely flyable. Of course, the lower DU was INOP, and the right RA was inop too, which meant the autopilot would disconnect at GS intercept...

Good thing two of us were there and could normally operate the aircraft without any problem, computers alone probably would have encountered a couple problems.

By the way, it is possible to fly manual CAT III approaches and landings, more than enough CRJs and others around that do it quite often.

[411A]

By the way, it is possible to fly manual CAT III approaches and landings, more than enough CRJs and others around that do it quite often.

Quite so, in the USA, this was pioneered by Alaska Airlines some time ago, using a HUD.
CATIIIA, I believe, and they are working on CATIIIB, so I'm told.
The HUD is very accurate.

[safetypee]

The HUD is very accurate,
but is the human? … and what about duplication, monitoring?
Using a single self-monitored HUD raises the question of how to monitor the human derived output. Use the other pilot and head down instruments, or with two HUDs, what if the humans disagree?
The automation issue is not what can be automated or at what cost; it is what should be automated and what reasons drive this decision. For the latter it is often the limits of human performance, the ability to conduct a task, or the reliability of the task in a range of conditions or over time.

[Flight Detent]

Are you ready to get aboard an airplane without a pilot...?

As mentioned earlier..."Not bloody likely!"

In fact I have some misgivings about getting on a long range airplane that doesn't have a professional Flight Engineer....:eek:

Cheers...FD...:\

[Massey1Bravo]

It will be interesting to see the career progression of today's young pilots over the next 40-50 years, assuming retirement age is raised to 70+. I seriously doubt the F/O position would still be viable at 2050-2060, and this would mean a mass cull of F/Os as the airlines renew their fleets in the future.

There's also the possibility of automation "dumbing down" the pilot's role to the point where the pilot becomes an extra flight attendant during cruise.

[Denti]

The HUD is very accurate,
but is the human? … and what about duplication, monitoring?
Using a single self-monitored HUD raises the question of how to monitor the human derived output. Use the other pilot and head down instruments, or with two HUDs, what if the humans disagree?
The automation issue is not what can be automated or at what cost; it is what should be automated and what reasons drive this decision. For the latter it is often the limits of human performance, the ability to conduct a task, or the reliability of the task in a range of conditions or over time.

Current installations usually have only one HUD on the captains side. Since the captain has to fly low vis landings anyway that kinda makes sense. Future HUD installation will be dual, for example it is standard equipment on the 787.

The fact is humans can fly precise enough to ensure landings nearly every time they try, the majority of all landings is done manually and only very few of them go astray. During low vis the issue is missing visual cues for a normal landing, however if those are artificially supplied by other means, in this case the HUD, humans are perfectly capable to their usual good job.

Sadly i did only my initial typerating on a HUD equipped 733 and haven't flown a HUD since, however during training with the HUD it was basicly a non-issue, the PM by default was the RHS and monitoring was not an issue, of course callouts if somethings doesn't fit, but that is something we do every day anyway, so nothing new there. If one of both calls a go-around it has to be performed, haven't flown in an airline that doesn't apply that principle in normal operation in the first place.