Automated Aircraft
 

Not exactly sure to post this one, but here goes. I'm at uni now, and one of lecturers (a former RAF engineer) is completely anti pilots and thinks the only reason they are on a flight deck is to drink coffee. The best thing we should do is just sit there and not touch anything. He believes that in a few years time (even though the technology is here now) all commercial jets will be completely computer operated with no pilot whatsoever.

I always try to rebut him by things like every single UAV has crashed, and pax wouldn't be confident to sit with only a puter there. etc. etc.

1) how do you feel about that??
2)how can i really rebut his statements before the summer hols start (after he's marked my exam paper of course) so he can sit and fume during what should be an otherwise great summer.:cool: :cool: :cool:

Thanks for your two pence worth:ok:

Hmmmm, I'm thinking of comments like "Why is it doing that?" and "What's it doing now?"

In various discussions I've had regarding full automation of both a flight deck (and ATC for that matter), often the topic of emergency situations and just exactly how much can be done remotely or automatically makes things interesting. IMO, all the variables that come into these unusual events require a couple of heads on the scene to ascertain exactly what's going on.

He sounds like he has a chip on his shoulder. Did he go through Cranwell looking for a pilot commission once??

In the very long term future we might see single pilot operation with one standby, and perhaps eventually in centuries to come, none. But I have to say that I'm an engineering student so understand the tech, and ignoring the fact that I want to fly I still would rather have a pilot sitting up there. And I'm an open-minded guy! It would take much time and adjustment for the market to build the sort of trust that customers must have in the system to loan their lives to something they associate with regular 'crashes'.

Andy

Ask him how often he reboots his PC due to human induced software errors or through situations that the software author didn't conceive. The principle is the same for automated systems even if the frequency is less.

Thats a great example Tinstaafl

I have to agree with whats been said - aircraft may still fly perfectly but in an emergency situation, IMHO they are screwed. The basic concepts of computing still say that a computer will only do what you tell it to. This brings in artificial intelligence, but this is hundreds of years away from anything near intelligent. One of the worlds leading AI chat bots oliverbot seems to work ok, until you ask it to explain one if its opinions or ask it a calculation.

Sorry if this has gone off track a little.

Regards

Maz

I remember reading that the flight control computers of the Airbus aircraft have logic pathways now so incredibly large that they cannot all be possibly checked in combination with themselves. Even simpler aircraft like the Boeing range can have their automatics doing very funny things such that you sit there saying 'what the hell's it doing?' before pressing the 'disengage' button and taking over manually to restore basic and human control over technology. Eventually the pilot will go, but not in our lifetimes! When the family in their shell suits are willing to climb into a pilotless aeroplane, our careers will be terminated, but persuading them that it is safe and sensible?.....Nahhh!

Tin, Maz,

"Ask him how often he reboots his PC due to human induced software errors or through situations that the software author didn't conceive."

Hardly a useful example. PC software is unreliable as to be otherwise would push the price up. Nobody's going to die as a result of your Powerpoint presentation going t1ts up.

"Ask him how often he reboots his FADEC/pacemaker/airbag/ABS software due to human-induced software errors." would be more useful examples. Truly reliable software tends to goes unnoticed because it never requires intervention by its operator.

The posts in this string reveal more about human nature than they do about automation.

The main obstacle to greater automation is fear of the unknown. The great drivers that will eventually bring it about are economics and safety.

We may worry about unreliable computers, but is it not the case that the majority of accidents are caused by human factors?

It may be true that it is impossible to test all of the software paths in a compex computer program, but when was the last time any of us had all of our mental pathways checked?

Full automatics will eventually happen. It is just a matter of when.
Then we can work on computerised passengers so nobody gets to fly.

True. So who do we get to program the computers, build and test the systems, and then install them?

Actually the last two posts are slightly wrong. Read a flight international just after newyears stating that in 2003 the greatest cause for accidents and loss of life (for the first time ever) was mechanical failure and not pilot error. Whether this is due to crappy maintenance or other problems i can't remeber, but thankfully CFIT is no longer the number one problem.

I have read the stats that say 80% or so of all accidents are from human error. That is comparatively easy to measure. Why? There's been a reportable incident or accident which is investigated.
What is much harder to measure is the number of situations that humans have sorted out before they have become accidents/incidents. By humans I mean engineers, ATC and pilots. Humans stop many incidents/accidents from happening.
Humans generally want to live to enjoy their day off.

Hmmm - of course it is possible to run the London Underground without drivers, but public opinion is against it.

It's happening with military aircraft right now. I can imagine unmanned freighters in 20-30 years, but passenger aircraft? Not for a while, unless we have more 9/11 events and need a truly sealed flightdeck

Had a problem with main gear not deploying on an elderly 737-400 the other night. Two green lights, one red. How would the computer open the inspection hatch and peer through to look whether the red markes were properly lined up?

IMO SLF wants to see something with a pulse up front driving. Automate the trolley dollies first.

I don't think we have a lot to fear yet. The extent of unmanned passenger vehicles seems to be airport transit trams and overhead monorails. When trains start becoming automated, then there is an indication our time is limited, but until then, I don't think there is anything to lie awake at night about. Automatic systems don't handle faults/fires/failures well. As for remote control- ask aeromodellers how many models they've lost with signal failures!

Hmm - the Docklands Light railway has been operating for years with driverless trains - without a major accident. And thousands of people use it happily every day.

The reason you don't see them on the undergound is more to do with the costs of converting 50 year old equipment and union resistance.


If you make it cheap enough they will climb in - happens in every other walk of life.

NotsoFantastic:

Docklands Light Railway has been around for years, but the differing scale of complexity makes trains and aircraft incomparable. For example, a train really has only one axis of motion to worry about, whilst an aircraft has three axes of rotation and then the actual position awareness.

Thought the example with the gear not deploying hit the nail on the head.

Andy

I cant. As a freight pilot in 20 years I will be flying aircraft that are being delivered today. We mainly buy 20+ year old aircraft.
The other point is GPS is going on 20 years old and we still cant use it for approaches, So why would UAV's have any shorter a gestation to commercial certification. Thirdly can UAV's fly non radar enviroment NDB to circling approaches to greek islands on changeover day. Er No.

Bit more work required methinks.

How will a computer judge a weather report,a required fuel figure based on assumed hold at lhr,or eventual night fog somewhere else.? Will you ,as a pax, trust a machine to avoid a cb,use the radar smartly to find best course through a ts area?
will a computer know that it should expect turbulance landing shortly after a heavy departure,or that those unsignificant long clouds can sometimes indicate severe turbulance,or...
I think a machine can outperform a human in many ways,but untill we'll have a cyborg on deck we'll see human pilots up in front.Even Enterprise had human pilots,excepting Data,of course.Which was one of a kind.
It is possible to build an automated plane even today,but my guess you won't find pax to fly with it.And , as Daysleeper said,no auto freighter too soon,either.
Tell your teacher that Airbus or whoever ,did built the automated plane.Will he be the first pasanger to try it,fly from Paris to Bangkok tomorow? Of course,with a fat insurance,just in case Bill mistaped some small software string.:ugh:
I won't.

"PC software is unreliable as to be otherwise would push the price up" - as a software engineer it is infuriating to read things like this. It is simply not true. For example, the Apache web server software, which runs pprune.org, ba.com, and so on, costs not a penny. However, as long as certain Large Software Companies can get away with selling unreliable crud, helped by statements such as the above, they will.

Apologies if this is out of place. I shall retreat to the SLF and computer forums again.

Daysleeper

Given the rapid developments in reconnaisance UAVs, the western need to minimise military casualities and the potential performance of unmanned combat aircraft, cruise missile technology (OK, so the landing's not up to much ;) ), it may be that on this timescale landings are handled by remote control. Maybe (even) by ATC.

I'd have thought that any new contemporary aircraft could easily be converted to autonomous and/or remote control given sufficiently robust computer systems, that I'm sure will evolve in my timescale. 25 years ago the concept of a personal computer looked a little far fetched, we've come a long way.

Cheers - SLF

Well on the surface a fair point, however, I'm flying 20 year old aircraft now. I've just converted from flying a 40 year old aircraft. Neither had any seriously major changes or upgrades to their systems in that time despite all the possible additions for efficiency and reliablilty. The cost of certifiying the changes would outweight the potential lifetime gain.
Aviation regulators , particularly in Europe, are conservative to the point of ludditism (or however thats spelled).
For instance and I know this is not directly relevent. the C-130 has been in RAF service for 30+ years. yet you could not certify one on the UK commercial registration.
Its all very well for the military to have UAV's but IMHO it will take a very long time before we see any in a public transport role.

ps , of course we could all be taking sub orbital flights to our holiday destinations by then. Go Spaceship One

http://www.scaled.com/projects/tiero...Index/body.htm

Daysleeper,

just curious, can't certify a C 130 on the civil register? why?

:confused:

FAA speak 1 10e9 In you and I speak this is pure bull crap.
 

The engineers can come up with the software and hardware that can cope with any conceivable situation however they can not come up with a design that will not fail. To do so the design of every single component must have a reliability of 1, which means it will never fail. This might be able to be accomplished through redundancy but this is also difficult to achieve.

Present FAA system failure requirements dictate that the system will not fail any more frequently than 1 10e9 or 1 billion hours of fleet operation. Some companies can manipulate the numbers to show that their systems will fail no more frequently than 1 10e17 yet these same aircraft suffer catastrophic failures resulting in loss of life and loss of the airframe.

In almost every civil aircraft design the airframe due to the stringent testing will show a reliability of 1 for the airframe and wings and even these design suffer catastrophic failures.

The engineers can design an aircraft that does not require human intervention and they can prove the reliability on paper showing that nothing will fail but ask them to be on the first test flight and most of them that do not have a death wish will find something else to do.


:E :E

mini

not 100% sure, my old company tried and failed but I was not directly involved.
I belive it was due to the performance with 2 engines out on one side

thanks for all the responses people. As always both sides of the argument are always welcome. I think i have enough info here to make a little speach just to prove my point. The looking through the hatch cause of the malfunctioning gear light is especially helpful. Oh and on the C-130 question, didn't atlantic use to run one from cov, before they decided on the electra. Or was that not on the UK register??

it will happen . maybe it will take 30 yrs ,but it will happen.only thing the autos cant do now are . they cant taxi the plane,they cant very sucessfully weather navigate, they arent the best at handling gusty winds that well [A/P on limitation is less than my actual limits].they cant take good decisions.
but all that can be fixed . they say today if you want to replace a human brain you need a computer the size of two newyork cities.
think about it . few years back a pc with 40mb was abig deal, today that damn thing fits in my pda.
gear problem .. it will have a system which can monitor it some other waylike in the NG they dont have viewers any more but another set of lights,so it can be fixed.
DONT GET ME WRONG i am a training captain with a major airline withe some 27 yrs to go. but my children will not be encouraged by me to take flying up as a career.
as far as things stand today .i will not step on a automated cocpit where the pilot only has to reboot the machine. wait a minute maybe they can do it from the ground........... happy landings

I doubt that the first examples will be totally pilot-less, its just that the pilot will be sitting in an office on the ground rather than at the pointy end at 30,000 ft.

So taxying, navigating around weather and making decisions about various faults are not that big a problem.

So what have we achieved? The source of human error is now on the ground and not first at the scene of the accident caused by his errors. It'll be no cheaper because the "safety critical" workers will now be engineers, flight ops or ATC who will demand higher salaries. All runways will have to be upgraded (see Greek Island thread).

UAVs have advantages because they do not have to be built with life support systems and their cargo is expendable. Doesn't apply to airliners.

There will never be an interim stage where only one pilot is carried "just in case". If you need one, you need two, in case one of them suffers "mechnical failure".

Just my thoughts on it.

Hi all,
IMHO,the discussion has put too much focus on the possibility of automation in one airplane.
Making an automated vehicle fly is one thing,operating it in the trtaffic environment we see to-day is altogether another proposition.
It will require a complete revamping of the ATC system,from equipment to procedures.And it will still require a back-up,i.e. a professional pilot to take over when the automatics are no longer enough.
So,if you only look at this subject from an economics viewpoint,it is not just feasible.

On a parallel subject,which is the fight against air terrorism,the SAFEE project (Safe Aircraft in the Future European Environment) is asking the right questions,for a very controversial subject,which is how to lock everybody on board a highjacked airliner -this includes the crew,of course- out of the controls,which will be then operated by a ground station up to a safe landing.SAGEM,BAE,Thales...are working on the project...
May be the subject deserves a discussion on the R&N forum.
Inputs,please!

I am sure that one day automation will replace the pilot, but there are many, many problems to overcome in the meantime.

One of the first questions that has to be asked is why you'd want to do it. Until it can be done more cheaply than it is now, with no loss of safety (or even an increase), then it won't happen. In the overall scheme of things, the cost of the two pilots is not that high - and there will be no manpower or salary savings at all if the pilots are replaced by ground-based operators working to European Working Time Directive limits, so pilotless aircraft will only make sense once all the decision-making and operating is done from within the aircraft by computers. This process must be absolutely robust and incapable of interference from unauthorised outside agencies.

Machine-based decision making is not good, generally, and needs to improve greatly before it can be relied upon to replace humans in the three-dimensional airborne environment. In fact, I could see automatic ATC coming before pilotless airliners! Radio-based ground-air communication is not currently secure enough to guarantee that pilot-free airliners could be controlled from the ground when necessary.

Comparing the civil and military environments is really not very helpful. Military UAVs exist because military pilots cost millions of $, and cannot be replaced quickly. The reliability and longevity of a military UAV is not very important, as they are built down to a minimum cost and in significant numbers. Civilian pilots are cheap, and can be replaced quickly, yet their machinery costs hundreds of millions of $ and must be incredibly reliable, and very long-lifed.

Aircraft coming into sevice now will be still around in 50 years time, or maybe more. No passenger-carrying aircraft are currently in development or proposed which do without a full compliment of pilots. That, to me, suggests that we are looking at a minimum of 70 years before any significant inroads are made into the pilot community by computers! I shan't worry about my job from that angle yet..

I'm not sure passenger aircraft will ever become fully pilot-less.
I'm sure that with future advances in computer technology it would be technologically feasible to do it... but would it be worth it.

Quite simply, the major advantages of current automated aircraft are:

1: they can do dangerous things in dangerous places without risking a life

2: they can avoid the need to build in complex and expensive life support systems, pressurisation, etc etc.

3: the aircraft can be built to a shape or design purely focused on its intended mission, without the compromise of adding bits to look after humans.

Airliners have passengers and the systems to support them.. pressurisation etc etc.
In my opinion, none of those 3 advantages of automation apply to passenger aircraft because

1: passenger planes dont generally do dangerous things in dangerous places

2: they need those life support systems

3: the mission of these aircraft is passengers

Ok, yes, some might argue that it may in time be cheaper to automate things than have humans do it. But I'm not convinced the degree of investment required in all parts of the system, from aiports to aircraft to ATC etc etc would be cheaper than a couple of pilots in the front.

LP

Scroggs,

"Machine-based decision making is not good, generally, and needs to improve greatly before it can be relied upon to replace humans in the three-dimensional airborne environment."

I would suggest the success of TCAS disproves this very general proposition about machines and the 3D airbourne environment. TCAS seems to have gained sufficient respect in its short life that its instant and reliable decisions are now preferred to those of ATC professionals.

I too don't believe that the pilot will be replaced overnight. There is nevertheless a growing army of little machines relentlessly making inroads into the decision-making prerogative of the pilot. Soon there will be no 'black art' or judgement left for pilots to claim as their own indispensable contribution.

tyro: " growing army of little machines relentlessly making inroads into the decision-making prerogative of the pilot"
name one :hmm:
actually ,excepting the tcas, there is no other system on a plane which will decide for the pilots or atc.Human authority prevails over all system on board.
There are some warning systems,but their destination is to improve the safety,alerting you of some erors you can do (for ex CFIT)
No system will decide instead of capt how to fly,plan a flight,solve an emergency (we have the QRH,but even there says ' checklists are not intended to replace good judgement.In some cases ,deviation from checklists may,at captain's discretion,be necessary.')
All systems will give inputs to the pilots,they decide what to do,based on sop's ,experience,good judgement,finally airmanship.Even on airbuses.
As I said before,even Enterprise,or Andromeda have pilots.Not to mention the 'positronic,...' brain. :ok:
My guess is that ,in some form or another,there will always be pilots on board of a plane.By our nature we like to be in control of our life.Not some 'friendly' machine,machine,ine,ine...
Brgds Alex

alexban,

Don't airbus aircraft computers prevent the pilot from flying outside the allowable flight envelope??

I do agree, the romantic notion that the pilot's "human authority prevails" is still commonplace in TV shows and movies.

Back in reality however, modern aircraft increasingly treat manual control input as requests rather than direct instructions to the hardware. There is a spectrum of machine intervention from a basic FADEC filtering out potentially damaging thottle settings, to the Airbus that intercepts and interprets attitude control inputs. These systems are programmed with their own ideas of what constitutes acceptable activity, ideas whose "authority" will always prevail over pilot input. To say that the pilot has ultimate control over these machines is, I think, practically meaningless.

As for solving emergencies, maybe these systems already pay their way by preventing many of them from happening in the first place.

I also agree with you that the desire to be in control is part of human nature. But so are human fallibility with repetitive tasks and in stressful situations.

There are lots of replies about safety here and that is of course the major principle. But safety is not just about 'not crashing'. It's about not getting into situations in the first place. It's the experience of a crew to avoid a TS upwind rather than downwind, to slow up on final if you can tell someone is dithering on the runway, to realise the most efficient way (which you can be damn sure the computers will be programmed with) is not always the best.

Pilots are here to stay. No computers can have experience and airmanship.

cortilla:
just curious,why would you want to fly outside of the allowable anvelope? Would you like 80 degrees bank? I guess not.
I was talking about flying,not doing aerobatics with a pax aircraft. :}

tyro : I don't have yet first hand experience on buses.But on Boeing,and other smaller planes I flew the pilot HAS ultimate control.
Don't talk now about pmc,or fadec ,or other system.This can fail and we have checklist and raw procedures for this case.
Yes,as I said to cortilla ,those systems will avoid exiting the allowable anvelope. On buses more than on other planes.But this is meant to help only.No plane will go on the GS,or climb.or whatever,by itself,with no command from the crew.
We are not talking here about keeping the plane running.We are talking about flying it.Of course the fadec will monitor the engint,avoiding high temp,etc.This is it's role.But it won't decide against a pilot who wants to shutdown the engine,for whatever reason.
We are talking about flying a plane in it's allowable anvelope.No plane will turn right if a pilot wants to turn left.
I don't know about your 'reality' but on mine,I think we won't see an airliner flown by computers.
Not saying it is not technically possible to make a an automated plane.The soviet shuttle Buran made a fully automated flight (supervised from the ground,i guess) But I think we won't see this for passanger transport. It is more than technical considerations regarding a normal,pax flight.
I am just curious,where did you get your ideas from.What autoplane you fly? Busdriver? Not even one will say the plane flies instead of him.
I hope we'll be around for the next 50 yrs :O and we'll see!
:ok:
brgds alex


ps hmm...., tyro ,are you an AI? just wondering...Is this Terminator 4?

Tyro TCAS is not a three-dimensional solution. It deals with three-dimensional problems in a two-dimensional way - ie it demands(it does not command) a vertical manouevre to avoid the perceived problem. It cannot demand a horizontal manouevre. Its thresholds are relatively generous, which is one of the reasons why a visual sighting of the 'threat' aircraft allows the pilot to override the demands of the system. But it's a good piece of kit, and probably the single most significant aid to flight safety introduced in the last fifty years,

However, it's a hell of a step to extrapolate the logic built into TCAS and extend it to the entirety of all the decisions that must be made in the execution of a long-haul scheduled passenger flight. There are far too many grey areas for any computer program to reliably and repeatably solve for it to be considered a safe way of conducting flights - for the moment. Believe me, I do this job for a living - with the aid of the best computers Airbus can give me. To put that in perspective, the computers in my brand-new Airbus are somewhat less capable than the Intel 486 I had on my desktop 10 years ago - and the software is mickey-mouse compared to FS2004! But it is reliable (more or less - and a lot more so than anything Microsoft) - and the passengers won't die if the software or hardware fails, because I and my colleagues are there to flexibly and appropriately apply our expertise.

It will come, but not in my professional lifetime.

There's a very thoughtful article on this topic here.

To summarise: pilotless aircraft are on their way...

Benet

(hopes pilotless helicopters may be further into the future!)

alexban,

I'm not saying that you'd want to fly outside the envelope as it were. you don't want to spill the tea afterall, i'm just saying that an Airbus will not allow you to do it. you say you want 40 degrees angle of bank (for whatever reason) and the bus will not let you do it full stop. Therefore the pilot does not have full control. The computer decides for you 25 degrees is the max allowable (not flown a bus before but i assume this is the max allowable), and there is no way you can override it. therefore in this situation the aircraft will decide against the pilot, and there is nothing that he/she can do about it.