NTSB update on Asiana 214

[Volume]

Absolutely incorrect. Fully automated robot aircraft, without expensive holes cut out all over the airframe, are just around the corner.

That only means the interface between man and machine will move, from the pointed end of the aircraft to an office where the machine is programmed or from where it is controlled. Human error can happen there as well, when programming the software and as well when using it (when programming the actual flight, just like using the FMS today). If the software interface for fully automated robot aircraft is poor, accidents due to issues in the man machine interface will continue. The only difference is, that the person doing the error will be among those who survive and can tell the tale...

[DozyWannabe]

@FGD135:

What you're not taking into account in your posts is that a lot of the examples cited are from decades before the introduction of glass cockpits, and would have thus been constrained by the limits of analogue technology on tape displays (which would in effect have been a large drum).

The Boeing research in the 1980s and the paper quoted does not take into account the possible bias inherent in the results due to most pilots of the day being more experienced with dials than tapes.

If you take these quotes into account:

Understanding the difference between that speed and current system state now requires the operator to perform mental mathematics, rather than directly seeing the difference.

Harris, 2004, noted, “the windowed design can be quite poor at providing the pilots with anticipatory information. On the electromechanical counter-counter altimeter, the altitude ‘bugs’ were always visible.” (p. 87).

a concept that is broken by (the common current solution) displaying important values numerically next to a graphic tape

and then look at the A330 PFD I posted earlier:



you can see that the bugs and relevant information are all in fact there, displayed graphically alongside the tape. In effect, the "proximity" perception is preserved in this manner (i.e. work from the yellow bug and avoid close proximity to the red [overspeed] and orange [alpha] bars - the lower red [stall] bars being off the bottom of the scale). The quotations above are inaccurate.

Originally Posted by ExSp33db1rd

On the Classic 747 there was a block of 20 engine instruments, 5 for each of 4 engines, in the centre instrument panel. With take-off power selected, all 20 'clocks' pointed to 4 o'clock, not be accident but by design. Should one parameter change, it stuck out like sore thumb. I can't comment on how this was dealt with on the 747-400 digital display, i.e. is it so patently obvious when something is wrong ?

I'd imagine that a noticeable EICAS warning would be generated, and hopefully tell the PNF exactly where the problem lay (in effect bypassing the previous method where the PNF would have to ask the FE what they thought the problem was).

[AGL_Guy]

After reading all this about automation of cockpits etc. and not looking out the big window in front, I ask m myself as an AGL Guy, why do we design and install PAPIs, Threshold Lights, Approach Lights, Runway Edge Lights and RWY Centre Line Lights if the Modern Pilots with all the smart stuff in their kit bags don't need them. Huge savings for the airport operator to leave them out.

But I suggest, there are a few fly boys out there who glance at these little luminaries from time to time, even if just to confirm that where they think they are is where they think they are.

[OK465]

....so that even if only a little extra focus is required on a tape, it is too much; cf recent accidents, and ASAGA.

safetypee,

the eye-tracker data in the referenced ASAGA study deserves comment in light of your comment about 'a little extra focus'.

Firstly, the two ASAGA simulator types (330 & 777) were both tape equipped. How do you derive the above conclusion without a comparison to eye-tracker data obtained in a round dial equipped simulator under the same circumstances.

Secondly, the percentage duration of eye-tracker time centered on airspeed by the PF on GA 1, 2 & 3 respectively were ~12%, 15% & 24%. GA 1 & 2 were hand flown, GA 3 was autopilot flown and one would expect that would allow for more time available to monitor anything and everything.

The average 777 runs were 67 seconds and the average 330 runs were 90 seconds. This equates to 8, 10 & 16 seconds for the T7 and 11, 13, 22 seconds for the 330. Are these times excessive and indicative of 'extra focus'? This is time the eye-tracker is centered on airspeed and not necessarily indicative of 'focus'. (Oh yes, we seem to be missing comparable round dial data)

If Asiana 214 had 'focused' on airspeed for any of these durations during the last 1-1 1/2 minutes of the flight we wouldn't be discussing relative merits of round dials versus tapes....or their apparent eye-tracker score of "0".

FGD135,

Good discussion, but you need to get into 447 threads and find out what that graph is really telling you about the speeds displayed and what you call 'spikes'. Computed Airspeeds are the values sent from the ADRs to the ASIs and NCD is No Computed Data and in that situation the tapes rest on around 46 KIAS (ADRs below 30) just like an electro-mechanical round dial pointer would with no air data input.

But that is "computed" airspeed, not necessarily the same thing that you would see if you were trying to display the pitot-static quantity on a big round dial.

It certainly is if the round dial is ADC fed. Otherwise....

I had pitot icing (rainy overnight and the required pitot system draining wasn't accomplished) in the last airplane I flew without a sophisticated ADC, the F-100. Climbing out of Sacramento, airspeed pointer dropped and pegged at the lower stop, flew pitch and power of course, and on that unusually cold day the nice folks at Nellis scrambled an F-111 to bring me down thru the wx on his wing. Told him to give me 180 KIAS on final and sure enough a little while after we got below the freezing level, the airspeed popped back to 180. If the plumbing gets blocked it doesn't matter whether ASI is fed by ADC or calibrated pitot pressure.

214 & 447 are not good examples to prove your psychophysics fueled point.

[BARKINGMAD]

Analogue keeps chimps happy.

Is all this stuff about modern instrumentation the reason why BA ordered their shiny 737-436s from Mr Boeing with big fat white needles behind glass, instead of the "new improved engine EFIS" criticised in the G-OBME Kegworth Accident report?


Let's rediscover the wheel again, spending lotsa money rounding off the edges of the hexagon.


Doesn't matter what the techno-geeks at the manufacturing end believe, we're still a primitive animal getting confused by all the shiny hi-tech gubbins which is installed with the aim of making our lives easier and maybe longer.


Back to the drawing board and try research on the human interaction with these sometimes user-unfriendly developments.


After the Strasbourg Airbus accident, it was suggested by some that the presence of a big fat VSI needle parked in the 5 o'clock position instead of the 7 0'clock mark, may have alerted that crew to their excessive RoD.

[safetypee]

OK465, whilst the discussion is about airspeed, the ASAGA study, and inference in this accident, suggests that it is the combination of factors which challenge pilot’s mental resources. In the extreme, and for a short period, the overall situation / task appeared to be beyond human capability. I doubt that any simulator experiment would accurately determine the combined effects; neither would a comparison with a dial equipped fight deck. In addition, the study involved more conventional approaches and automation which worked as imagined, … up to the point of GA.

All that might be deduced about the speed display in this accident is that it is possible that the tape display contributed higher mental workload than might have a dial instrument, based on background research, but what the effect was in comparison to the mental effort in understanding the state of automation or aircraft flightpath cannot, and should not be judged.

Other aspects with potential for greater influence could arise from management policies. It is reported that the operators automation policy was to maximise the use of automation, which could be interpreted in three ways: AP+AT, Man pilot+AT, and Man thrust+AP, but there was no specific clarification or advice. Yet the excluded combination Man pilot+Man thrust, if used could have had significant influence on the approach and crew performance.

Also, aspects of the report could be interpreted as the training policy / economics only allowing (allocating) one attempt at each training scenario, thus both the trainee and instructor may have expected to succeed at all times. If this view influenced the crew then there could have been greater pressure on the trainee to resolve the situation, and for the inexperienced instructor not to fail a candidate (a form of press-on-itis). Whereas the situation (from hindsight) actually required early intervention, which without judging pass/fail or allocating blame, was an ideal learning opportunity (experience) from the debrief.

As with airspeed, the contribution of policy cannot be judged, but these should be considered under ‘what if’, inviting individuals and organisations to look at the details or interpretations of their policies and procedures.

[DozyWannabe]

Originally Posted by BARKINGMAD

Is all this stuff about modern instrumentation the reason why BA ordered their shiny 737-436s from Mr Boeing with big fat white needles behind glass, instead of the "new improved engine EFIS" criticised in the G-OBME Kegworth Accident report?

Which sidesteps the fact that the other reason the gauges were misread in the Kegworth accident was that the analogue dials used to display engine vibration on the -300 series were considered notoriously unreliable and difficult to read, whereas Boeing's electronic replacements were simply difficult to read.

[AirRabbit]

Originally Posted by AGL_Guy

After reading all this about automation of cockpits etc. and not looking out the big window in front, I ask m myself as an AGL Guy, why do we design and install PAPIs, Threshold Lights, Approach Lights, Runway Edge Lights and RWY Centre Line Lights if the Modern Pilots with all the smart stuff in their kit bags don't need them. Huge savings for the airport operator to leave them out.

But I suggest, there are a few fly boys out there who glance at these little luminaries from time to time, even if just to confirm that where they think they are is where they think they are.

Thanks for opening this particular door … as what I’m about to say is very likely the completion of the thoughts you’ve initiated with your comments …

Actually, pilots should be using ALL of those “little luminaries” together WITH ALL of what you’ve labeled “smart stuff” (presuming you mean all the instrument presentations currently available – where ever and how ever they are presented in any specific airplane) and using all of this information ALL OF THE TIME – including CAVU at high noon! The more information pointing to the current position and status of the airplane that the pilot can and does reference, the more that pilot will be able to select those pieces of information on which he/she feels most comfortable in using – AND referencing - ALL of that information, ALL of the time, allows the pilot to see how activation of the controls affects the resulting position and condition of the airplane, as that position and condition is displayed in ALL of those pieces of information … collectively and individually … in effect, providing a training exercise each and every time he/she brings the airplane in for a landing. If any current pilot does NOT know and recognize these basic aspects, then the instructors they have had have not done their job as well as they should have (and remember that instructors have a regular and recurring opportunity to ensure that the pilots they train are, in fact, trained and competent aviators) – and those pilots who have been "cheated" on their training are, in turn, increasingly likely to become involved in an accident or incident – or at the very least, in a personal crisis that may be known only to him/her – and hopefully THAT will all it will be!

I, for one, have stood at the edges of far too many “smoking holes” that should have never existed - and might NOT have occurred if the pilot had a better understanding of what the airplane was doing and what he/she should have done to correct it!

[ExSp33db1rd]

............I'd imagine that a noticeable EICAS warning would be
generated..........

How ? Just a word on a display that one is not looking at at that time ? or maybe a warning light? maybe a horn? whatever, the eyes have to be taken off the job in hand and a computer screen scrutinised, buttons pressed, information read, understood and processed.

Wot's esier than noticing a problem out of the corner of the eye and KNOWING INSTANTLY the problem whilst still controlling the task in hand?

Apologies for reptition, but I've mentioned before that some of the WW II Captains I started flying with had trouble ( some of them ) flying a rudimentary instrument approach, even just a simple NDB approach, but pop out of cloud at minimums, too fast, too high, no flaps etc. and say "The runway's over there, Sir" (never forgetting the Sir ! ) and they would straighten up and fly an immaculate visual approach to a perfect manual touchdown. 40 years later I was teaching co-pilots, brought up on a diet of Space Invaders and Bill Gates' latest toy, who would fly an instrument let-down better that I ever could, or would, but pop out of cloud at minimum, in the slot, configured for landing,stabilised but then couldn't put a real aircraft on to the real Earth without assistance.

Somewhere we've gone wrong in the intervening years.

[safetypee]

AR, if we consider your views without judgement, this might help us to identify alternatives to the idealist situation described.

We are experts in what crews ‘should’ have done after the fact, and how to use this distorted knowledge for futuristic solutions. Alternatively we might consider what the crew ‘could’ have done in the particular circumstances, what were they actually capable of, and thus what insight this gives us.
The mantra of “use everything - all available resources”, is continually repeated, more often without thought as to what is meant. Is it possible to use everything, what is ‘everything’, does it vary with situation, training, experience?
Does the industry teach what aspects are required in each situation, how to identify what information to use (data with understanding), vice those cues we might feel comfortable with. Is all information the same, what is important and why, how are these aspects referenced, prioritised, rated, individually and particularly collectively?

How might we know if pilots do not recognise appropriate factors? There may have been many successes in training and previous operations, but encountering an unusual situation then the application of previous skills could fall short of the performance required for any number of human factors reasons.

How do pilots acquire that ‘understanding’ which is so easy to identify with hindsight; yet the industry continues to churn out mantras full of assumption, and perhaps without the necessary thought and explanation to ensure understanding.
We cannot use ‘all of the information all of the time’ (a limit of human performance); but we might use an adaption of the rest of Abraham Lincoln’s quote; perhaps there is something to be learnt from that.

The industry appears to be blinded by hindsight, attempting to find patterns and trying to resolve past problems with the same fixes, resulting in few real gains in improving safety – how can we move forward by looking backwards.

[FGD135]

What you're not taking into account in your posts is that a lot of the examples cited are from decades before the introduction of glass cockpits ...

Dozy, that makes no difference. In fact, it makes no difference whether the human subjects are pilots or not - all humans will have the same problems with tapes. This is because of how our brains work.

The Boeing research in the 1980s and the paper quoted does not take into account the possible bias inherent in the results due to most pilots of the day being more experienced with dials than tapes.

Do you actually know how the Boeing research was conducted? You make it sound like they just asked a couple of old-timers for their opinions. Here is that bit about Boeing again:

The Boeing Company conducted extensive research in the mid 1980’s into vertical tape instruments, finding some concerns:

They lacked relationships that were used extensively by pilots in performing flight tasks. This perception was strengthened by human factors research, which also concluded that, in general, moving scale displays are not as effective as moving pointer displays.

Driven by explicit airline demands for the maintenance savings of CRTs over electromechanical pointers and the space requirements of matching the Airbus eight-inch screens, Boeing eventually chose vertical tapes for the 747-400.

I'd imagine that a noticeable EICAS warning would be generated, and hopefully tell the PNF exactly where the problem lay ...

That would be a poor solution - which would also make the cockpit just that little bit more complex.


20 needles that should all be pointing in the same direction, combined with a human eye and experience, is to me the most optimum relationship that we can have between man and machine. From just a glance the pilot could see that 20 big and important parameters were normal - can it get any better than that?


Some of those needles might have been dancing around or jittering, but the human is so good at filtering such observations that that would have made no difference to the big picture. To a computer, however, jitter or noise on the parameters may give rise to nuisance warnings.


When instrumentation was using the big pointers and making use of the human sensitivity to parallelism we were at the pinnacle in cockpit ergonomics, in my opinion. Since then, we have gone backwards.

After the Strasbourg Airbus accident, it was suggested by some that the presence of a big fat VSI needle parked in the 5 o'clock position instead of the 7 0'clock mark, may have alerted that crew to their excessive RoD.

Excellent point, BARKINGMAD. I have been meaning to comment on the VS presentation as I believe that too is woefully deficient. VS is not as important as airspeed, hence my silence until now.

I agree that, had that crew had the big VS dial of yesteryear, they may well have noticed that something was not quite right.

Look at that picture of the A330 PFD that Dozy posted. I have complained about the size of the tiny patch in the middle of the airspeed tape before, but the VS indication area is even more tiny! (On that PFD design, anyway).


OK465, do you know exactly how the A330 arrives at "computed airspeed"? Please enlighten me if you do. If the aircraft had had the big round dial from day 1, then the way the airspeed was computed may not have been the same as it was for the PFD. Also, to save me delving into the report, which ADC was feeding the FDR?

... with a press of a couple of buttons ( if I can remember which ! ) change the 'picture' to a digitalised display of 'round' instruments, to increase my comfort zone.

ExSp33db1rd, you and the round dials are a good fit. Not because you are an 'old fart', but because you are a human being. Keep using them. You and your passengers are safer for it, and we now know that there is some scientific support for this conclusion!

[porterhouse]

and we now know that there is some scientific support for this conclusion!

there is none regardless how many times you will write it.

[Volume]

Actually, pilots should be using ALL of those “little luminaries” together WITH ALL of what you’ve labeled “smart stuff” (presuming you mean all the instrument presentations currently available – where ever and how ever they are presented in any specific airplane) and using all of this information ALL OF THE TIME – including CAVU at high noon!

This is a dream and will probably never happen. If you have redundant information available, it is quite normal that you concentrate on the most convenient one and ignore the others. It takes a lot of training and discipline to make people use redundancies.
If you want redundant information to be noticed, it must "jump" at you. One light of a PAPI changing from white to red might indeed draw more attention, than an indicator moving constantly and slowly along a scale from the white into the red. Hence digital information (and a PAPI is nothing else, although older than the term itself) might be harder to understand when you try to monitor it, but easier to recognize if it changes.
With respect to this, the PFD shown above will draw more attention when the first 3 altimeter digits "jumps" from 400 to 399, compared to the speed tape numbers just constantly and slowly shifting vertically. If you want changes to be noticed, the indication must change in a very obvious way.
If a change in speed indication would obviously "jump" every 5 kts (or make a sound passing a round 5 kts number, different for increasing and decreasing, just like the clicking noise old fashioned electromechanical instruments made), it would be much more obvious to the pilot than a dial passing a certain position (unless that position is a very distinctive one, e.g. the 9 ´o Clock position)
An indication designed for quick and accurate reading, an indication designed to se a trend, and an indication designed to draw attention when changing must be designed differently.

So I think the first thing we must agree on is what a pilot in a highly automated aircraft should do about airspeed. Monitor it? Manage it? Track it? Notice it if it changes? Notice if it is too low? This all required different design of the instrumentation for optimum performance.

[oblivia]

Hmmm … it must be nice to have such an infallible “crystal ball” at your elbow. Personally, I have no idea about what the next 20 – 50 years of aviation is going to see. Could it include “fully automated airplanes” ?? … sure … it could but if that does happen I would suspect that the cost of an airline ticket might preclude all but the wealthiest of persons from traveling by air.

The history of commercial aviation to date is one of increasing automation, reduced cost and reduced real incomes for pilots. I don't think these things are unconnected and see no reason why further levels of automation would produce an opposite trend.

The costs of processing power and wireless communication are only going in one direction. And computers don't need rest breaks, hotel rooms, sick leave or holidays, don't sue for wrongful dismissal, don't ask for pay rises and are already responsible for fewer mistakes than humans. They would probably also mean lower insurance premiums.

IBM's Watson is already better at medical diagnosis than the average doctor — and that's not based on it being given a data set; it speaks to patients. Is flying a plane vastly more difficult than medical diagnosis? And yes, I realise the challenges are different, but they're still just computational.

Indeed, the obstacles probably aren't even technological at this point. The obstacles are pilots and passengers, neither of whom are massively keen on the idea. But it will be passengers who make the final decision — and they will vote with their wallets, as they've done throughout the history of aviation.

And if we don't do it, the Chinese will — because they won't care about firing pilots and their passengers only care about price, not least because their mistrust of humans is ingrained by decades of experience at the hands of incompetent and corrupt institutions.

Indeed, the industry's growth is not going to be in Europe or the US — it's all in poor countries where passenger growth can be in the double digits. Where will all the pilots come from? Will they be adequately trained? Would you be so sceptical of automation if you lived in Bangladesh or Uganda?

[Volume]

Is flying a plane vastly more difficult than medical diagnosis?

It is. Reaction time is essential, timing is crucial. In decission making without time contraint it is hard to beat a computer. In reacting to a surprising situation never thought of, it is hard to beat humans.
However, computers in airplanes are still very, very basic using very simple algorithms and laws. There is no doubt that a computer could do many jobs much more accurately than a human. However, the more complex you make it, the more bugs can remain hidden until they are discovered by accident (by an accident or after an accident more probably...). It is also much easier to install true redudancy with humans (they are all different, except for one in the brian movie ;-) than with computers, where preventing commom mode failures is really rocket science.

[AirRabbit]

Hi Volume: It seems to me that you’ve outlined the necessities of improving the training that ALL pilots really should be provided – even though I think you were headed down a somewhat different path, what you said can easily be focused on the necessity of reviewing, revamping, and fortifying the training we provide to all pilots … but particularly to airline pilots.

Originally Posted by Volume

If you have redundant information available, it is quite normal that you concentrate on the most convenient one and ignore the others. It takes a lot of training and discipline to make people use redundancies.

I completely agree … ALL of aviation requires “…a lot of training and discipline…” hence, my recommendation – nee, my urgent recommendation – that we (as participants in an ever-changing industry) MUST receive the necessary training so as to develop and understand what we see, hear, and feel, so as to recognize and use the displays provided and the information presented.

Originally Posted by Volume

If you want redundant information to be noticed, it must "jump" at you.

Certainly, when redundant information does indeed “jump” from one indication to another, that change in the display is likely to draw some attention. The issue is to train the user to notice that display – even if it doesn’t change, but certainly if it does … even a minor amount.

Originally Posted by Volume

One light of a PAPI changing from white to red might indeed draw more attention, than an indicator moving constantly and slowly along a scale from the white into the red. Hence digital information (and a PAPI is nothing else, although older than the term itself) might be harder to understand when you try to monitor it, but easier to recognize if it changes.

First: IF the brightness of that PAPI light is such that any change will garner the attention of the pilot (who [i]should]/i] be busy monitoring ALL of the indications and controlling the airplane) it is very likely to be noticed. HOWEVER, such brightness is not typically present until and unless the distance to the PAPI lights is relatively short and the weather conditions are such that there is little if any obscuration, and, even then, particularly if the airplane is further out on final approach, if the pilot is focusing on some other display or indication where a majority of that pilot’s attention is being focused on that “other” issue (whatever it may be), such a change in the PAPI display may only be recognized when/if the pilot’s attention is again returned to the PAPI display and the pilot recognizes that the display now being seen is different from what was seen when it was last observed and recognized. “Seeing” something is not necessarily the same thing as understanding the meaning of what is being seen (particularly so when the current display is different from what was displayed previously). Additionally, pilots are expected to, and must be able to, see, hear, note, analyze, decide, and react to the numerous information sources we all know is present in an airplane cockpit – at any time – and particularly during an approach … either visual or instrument.

Second: It should be obvious that “…an indicator moving constantly and slowly along a scale…” should warrant an increased amount of awareness of that movement, even if the movement is desired. Is it moving in the correct direction? Is it moving at an acceptable rate? Can that movement be immediately stopped/reversed, if necessary or desired? How do you know?

Originally Posted by Volume

If you want changes to be noticed, the indication must change in a very obvious way. If a change in speed indication would obviously "jump" every 5 kts (or make a sound passing a round 5 kts number, different for increasing and decreasing, just like the clicking noise old fashioned electromechanical instruments made), it would be much more obvious to the pilot than a dial passing a certain position (unless that position is a very distinctive one, e.g. the 9 ´o Clock position).

The rate of change of most things of interest to a pilot when flying an airplane (particularly when on final approach), is critical; but if it were true that such changes would be noted ONLY through some “obvious, attention grabbing manner,” it would be necessary to provide a bombarding cacophony of new sounds, changes of sound, flashing lights, lights changing colors, and whatever myriad of other “immediate attention grabbing” features that might be considered “obvious.” In the alternative, I seriously believe that the recognition of “changes” has to be achieved through constant vigilance and constant awareness on the part of the pilot flying the airplane. Some of those changes will be anticipated and be recognized as supportive, and some very well may be indicative of something other than what is desired, and may require an immediate intervention/correction by the pilot. Additionally, the pilot monitoring should ALSO maintain that same amount of vigilance and constant awareness – temporarily interrupted only by necessary checklist requirements or radio calls – and that is both temporarily (not continuously) and then only interrupted (meaning that the conscious effort regularly returns to “vigilance” ASAP). However, this kind of vigilance and needed recognition cannot be achieved in the absence of determined, directed, and meaningful training.

Originally Posted by Volume

An indication designed for quick and accurate reading, an indication designed to set a trend, and an indication designed to draw attention when changing must be designed differently.

I’m not at all sure that the “different” design has not already been achieved and employed. What I’m reading here is, in my opinion, a description of every flight instrument in every airplane.

Originally Posted by Volume

So I think the first thing we must agree on is what a pilot in a highly automated aircraft should do about airspeed. Monitor it? Manage it? Track it? Notice it if it changes? Notice if it is too low? This all required different design of the instrumentation for optimum performance.

Actually, ALL of these things are required! ALL the time! ALL related to ALL the parameters involved in airplane flying! Airspeed, altitude, rate of climb, rate of descent, heading, pitch attitude, bank attitude, coordination, engine parameters, airplane configuration, and on and on and on……..

Originally Posted by Volume

Reaction time is essential, timing is crucial.

In a great many areas – and very easily, a majority of areas – this is certainly true.

Originally Posted by Volume

In decision making without time constraint it is hard to beat a computer.

Actually, computers do not “make decisions.” They do one of 2 things: 1, (passively) they present information that has been sensed; and 2, (actively) they compare parameters and, depending on the peripherals involved, make adjustments to settings as have been previously programmed, until what is sensed matches what has been programmed.

Originally Posted by Volume

In reacting to a surprising situation never thought of, it is hard to beat humans.

I would agree, whole heartedly … but ONLY when the human(s) we are describing has/have been thoroughly, properly, and completely trained. An unprepared human is very much like a computer without an operating program … the lights may be “on,” but rarely is “anyone home.”

[AirRabbit]

Originally Posted by oblivia

The history of commercial aviation to date is one of increasing automation, reduced cost and reduced real incomes for pilots. I don't think these things are unconnected and see no reason why further levels of automation would produce an opposite trend.

The costs of processing power and wireless communication are only going in one direction. And computers don't need rest breaks, hotel rooms, sick leave or holidays, don't sue for wrongful dismissal, don't ask for pay rises and are already responsible for fewer mistakes than humans. They would probably also mean lower insurance premiums.

Of course, if we use computers aboard airplanes to do what pilots do today … the uninformed may well conclude that the direct operating costs would decrease. The point I was making in my earlier post on this issue did not address pilots being replaced with computers (more on that in a moment) … I was addressing the point brought out by Chronus when he posted…

The airborne pilot is now basically a computer monitor, not long before taking a seat alongside his ATC counterpart.

…in which he was clearly referencing the on-board pilot being replaced with an “off-board” pilot (taking his/her seat next to his Air Traffic Controller counterpart) flying the airplane through air-ground “telemetry” … much like the “UAVs” drawing such public and private rumblings in current news stories. It was on this basis that I made the statement that the costs involved would likely soar out of sight and the only recouping of such costs would likely fall to an increase in airline ticket prices – and, while I’m not an expert at such economically driven issues (an woe were that to ever occur!) but it seems logical that this would be the most direct and meaningful way to disperse that increased cost.

Originally Posted by oblivia

IBM's Watson is already better at medical diagnosis than the average doctor — and that's not based on it being given a data set; it speaks to patients. Is flying a plane vastly more difficult than medical diagnosis? And yes, I realise the challenges are different, but they're still just computational.
Indeed, the obstacles probably aren't even technological at this point. The obstacles are pilots and passengers, neither of whom are massively keen on the idea. But it will be passengers who make the final decision — and they will vote with their wallets, as they've done throughout the history of aviation.

Not being a medical doctor, I’m probably not the best person to determine if flying a plane is vastly more difficult than completing medical diagnosis. But, to get to the full effect of such an analogy, all of the difficulties in flying an airplane should not, in my not-so-humble opinion be compared to medical diagnosis, but, rather, to actually performing the operation for which the diagnosis was indicating is necessary.

However, all of that not-with-standing, I am of the opinion that when medical patients, with the full confidence and agreement of the families of those patients, are willingly wheeled into an operating room, completely devoid of medical staff (doctors and nurses) and are, instead, staffed with the most current version of computer-run robotics, complete with anesthetics, antiseptics, and other drugs, scalpels, saws, drills, clamps, swabs, sutures, and all the other paraphernalia that are typically used/required in even the most basic of surgical operations – all of which will be under the control of, and actually wielded by, a preprogrammed computer, or a bank of such computers – I think we may have reached the point where some healthy persons may think it appropriately safe to place their lives in the hands of a preprogrammed computer flying from one point to another point on the globe – but until that medical scene becomes reality, I think the airplane load of passengers at the mercy of an autonomous computer is still some distance in the future.

As I understand the current “state of the art” in what I believe is referred to as “robotic surgery,” the use of “robotic systems” is limited to aiding in surgical procedures where the surgeon uses one of several methods to control the instruments; either through some kind of direct manipulator or in rare situations, through computer control. Again, as I understand it, the primary benefit of using such “robotic capabilities” is in performing certain actions with much smoother, controlled motions than could be achieved by a human hand. The main object is to reduce or eliminate the tissue trauma traditionally associated with open surgery without requiring more than a few minutes' training on the part of surgeons. All of which are controlled by and under the direct participation of the surgeon. This sounds to me like a pilot using an “autopilot” to fly the last portion of an instrument approach procedure … and the very advanced systems actually land the airplane at the conclusion of the approach. Pretty smart stuff, I’d say!

Originally Posted by oblivia

And if we don't do it, the Chinese will — because they won't care about firing pilots and their passengers only care about price, not least because their mistrust of humans is ingrained by decades of experience at the hands of incompetent and corrupt institutions.

I’m not at all sure about the accuracy of this opinion … however, … if someone with a similar disregard for the potential problems were to determine what part of the world would be most suitable to introduce such a program, it might just be thought that a section of the world that could most likely endure the potential sacrifice of a good portion of its population to verify the viability or confirm the absurdity of such a practice, perhaps those countries with burgeoning populations might prove to be the most logical to attempt such a feat. I sincerely believe that the pivotal aspect of such an enterprise would not be the hiring or firing of airline employees … but the potential longevity of its customer base might be critical in such business decisions.

Originally Posted by oblivia

Indeed, the industry's growth is not going to be in Europe or the US — it's all in poor countries where passenger growth can be in the double digits. Where will all the pilots come from? Will they be adequately trained? Would you be so sceptical of automation if you lived in Bangladesh or Uganda?

I’m not at all sure that my position is correctly described as being “skeptical of automation,” and I have virtually NO knowledge of what level of skepticism currently exists, toward anything, among the populations in those countries. Personally, I am a very big proponent of properly designed and properly employed automation – and I believe we haven’t yet seen the end of improvements and capabilities. Additionally, I would suspect that the younger citizens of the kinds of countries you reference are likely not as “current” with the existing status of automation, nor its potential problems, solutions, or applications as some younger citizens are in other parts of the world, but I am at a loss as to how that might affect the design, development, deployment, or anticipated problem areas with any current or future capability or desire for the application of automation.

[Chronus]

AirRabbit says

I think the airplane load of passengers at the mercy of an autonomous computer is still some distance in the future.

But work in the field of full automation is well underway. The case in point is the ASTREA PROJECT which is for the use of UAS in all classes of airspace without the need of special or restrictive conditions of operation.

Surely a project of this magnitude is not without a specific aim and purpose.

[porterhouse]

But work in the field of full automation is well underway

In terms what is feasible in the near future in passenger aircraft it is not an unmanned aircraft but SOP - single pilot operations. In other words an aircraft cockpit with single pilot only. There would be another "virtual" pilot on the ground. NASA is actually in the midst of extensive studies/simulations of this very topic. I work at NASA Ames and this is the hottest topic pursued at the moment in the aeronautics division of NASA. Air Rabbit is correct - a fully autonomous passenger aircraft is not yet on anybody's drawing boards, but SOP is very much on the drawing boards.

[AirRabbit]

Originally Posted by Chronus

AirRabbit says

I think the airplane load of passengers at the mercy of an autonomous computer is still some distance in the future.

But work in the field of full automation is well underway. The case in point is the ASTREA PROJECT which is for the use of UAS in all classes of airspace without the need of special or restrictive conditions of operation.

Surely a project of this magnitude is not without a specific aim and purpose.

I am well aware of the UAS efforts underway in the US – in fact I have several colleagues who are intimately involved on both sides of the regulatory fence on this very issue. But, I say again, in each of these situations, one of the serious issues facing the operation is the connectivity between the “pilot” on the ground and the airborne UAS – which presents its own interesting set of problems and solutions. In fact expanding the existing UAS program into a world-wide airline operation is not only not in a similar neighborhood, it’s not even on the same or a similar planet – something a lot closer would be playing with a 4-piece, slip-together, balsa-wood airplane in preparation for operating the space shuttle replacement.