Automation dependency stripped of political correctness.
Check Airman
Depends on the fleet - it is certainly the case on the 777, I believe that's also the SOP on the Airbus fleet,
It is not a no-go item.
No.
Please note that I am not saying whether I am a supporter of the policy...or not...
A previous poster mentioned that BA doesn't allow the AT to be disconnected. Can anyone confirm that?
Is the AT considered a no-go item?
Is it an emergency if the AT fails in flight?
Please note that I am not saying whether I am a supporter of the policy...or not...
Join Date: Mar 2006
Location: USA
Posts: 2,513
Likes: 0
Received 0 Likes
on
0 Posts
Thanks Wiggy.
One really has to wonder about the competence of manual writer(s).
You have a system that you're never allowed to control manually, but you may dispatch with the auto mode inoperative, and are expected to fly the plane to the same level of precision- a task you have never attempted.
I don't know of any US carriers with that sort of automation policy. Does anyone have any theories on why European carriers seem to take these extreme policies?
One really has to wonder about the competence of manual writer(s).
You have a system that you're never allowed to control manually, but you may dispatch with the auto mode inoperative, and are expected to fly the plane to the same level of precision- a task you have never attempted.
I don't know of any US carriers with that sort of automation policy. Does anyone have any theories on why European carriers seem to take these extreme policies?
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
I'm getting really bored of answering this question.
Yes, It could.
NASA trialled a system in the 70's that dealt with it so well that a PPL managed to land the aircraft.
It was transparent to the pilot and took about 2 seconds to fully learn how to fly using only variable thrust.
Re can computers make decisions like humans.
No, they don't get scared or excited and make errors.
They think at 1000 times the speed and get all the maths right. They never misread a landing distance chart, they know which is the nearest suitable airfield.
The one area that a human will be better than a human for the near future is black swan event.
There are very few of them.
There are a lot of pilot errors.
That sort of thing is exactly what computers are really good at.
Can't be bothered to hunt for the links again.
Yes, It could.
NASA trialled a system in the 70's that dealt with it so well that a PPL managed to land the aircraft.
It was transparent to the pilot and took about 2 seconds to fully learn how to fly using only variable thrust.
Re can computers make decisions like humans.
No, they don't get scared or excited and make errors.
They think at 1000 times the speed and get all the maths right. They never misread a landing distance chart, they know which is the nearest suitable airfield.
The one area that a human will be better than a human for the near future is black swan event.
There are very few of them.
There are a lot of pilot errors.
That sort of thing is exactly what computers are really good at.
Can't be bothered to hunt for the links again.
Computers are built and programmed by humans.
Computers cannot think.
A computer will literally do exactly what it is told. If it has been misprogrammed to fly towards a cliff, or below MSA then it will do so.
Errors are made, databases have been wrong.
Some of these errors are not apparent during testing and validation. They only appear one day when an unforeseen unique set of circumstances arise.
NASA's Apollo 11 LEM computer was taking it towards a boulder field to land, until the human pilot, a certain Mr Armstrong, took manual control and vectored it clear.
In the event of failures of aircraft systems, the computer(s) may not have been programmed to respond appropriately, or may be incapable of responding correctly owing to loss of hydraulics or electrics or atmospheric measurement. That's why you always need humans in the loop.
Humans are not infallible of course. This is why you need two of them, and proper, thorough training, proper experience, and properly rested and motivated crews. Today's economics has taken precedent over this and we are seeing the results - a series of crashes and incidents that should never have happened.
I think it is significant that these 'automation dependant' crashes have started occuring now that pilots go from small training piston aircraft straight onto modern jets, missing out what used to happen - which was a few year's apprenticeship on turbo props, actually flying the aircraft and learning the ropes in a commercial environment, and applying all the things that had been learned in the ground school.
Computers cannot think.
A computer will literally do exactly what it is told. If it has been misprogrammed to fly towards a cliff, or below MSA then it will do so.
Errors are made, databases have been wrong.
Some of these errors are not apparent during testing and validation. They only appear one day when an unforeseen unique set of circumstances arise.
NASA's Apollo 11 LEM computer was taking it towards a boulder field to land, until the human pilot, a certain Mr Armstrong, took manual control and vectored it clear.
In the event of failures of aircraft systems, the computer(s) may not have been programmed to respond appropriately, or may be incapable of responding correctly owing to loss of hydraulics or electrics or atmospheric measurement. That's why you always need humans in the loop.
Humans are not infallible of course. This is why you need two of them, and proper, thorough training, proper experience, and properly rested and motivated crews. Today's economics has taken precedent over this and we are seeing the results - a series of crashes and incidents that should never have happened.
I think it is significant that these 'automation dependant' crashes have started occuring now that pilots go from small training piston aircraft straight onto modern jets, missing out what used to happen - which was a few year's apprenticeship on turbo props, actually flying the aircraft and learning the ropes in a commercial environment, and applying all the things that had been learned in the ground school.
What defines the nearest suitable airfield then?
It's not just the nearest airfield that the aircraft can safely land at.
It depends on a whole host of different factors.
In a medical emergency one airport further away may actually be the quicker option for the necessary hospital care.
You're right, humans do make mistakes. Thats what cross-checking is for, and it does a good (not perfect) job of eliminating or identifying as many mistakes as possible.
But a computer is programmed not by a computer but by a human. They can make mistakes too.
It's not just the nearest airfield that the aircraft can safely land at.
It depends on a whole host of different factors.
In a medical emergency one airport further away may actually be the quicker option for the necessary hospital care.
You're right, humans do make mistakes. Thats what cross-checking is for, and it does a good (not perfect) job of eliminating or identifying as many mistakes as possible.
But a computer is programmed not by a computer but by a human. They can make mistakes too.
Join Date: Feb 2015
Location: Alternate places
Age: 76
Posts: 97
Likes: 0
Received 0 Likes
on
0 Posts
LlamaFarmer, to your point, I was googling the abbreviations in your post, (DODAR, etc) and in the list of hits was this AAIB Report on two UAS Events, (same aircraft).
Like QF32 and numerous others I can recollect, including some from personal experience, I think it is a useful if not familiar example of why autonomous flight is a very long way from a successful, demonstrably lower risk, commercial implementation, ("lower-risk" being the erstwhile* raison d'être behind the notion of autonomous, routine, commercial flight).
*erstwhile, because I don't believe the primary reason for autonomous flight is risk-reduction
Like QF32 and numerous others I can recollect, including some from personal experience, I think it is a useful if not familiar example of why autonomous flight is a very long way from a successful, demonstrably lower risk, commercial implementation, ("lower-risk" being the erstwhile* raison d'être behind the notion of autonomous, routine, commercial flight).
*erstwhile, because I don't believe the primary reason for autonomous flight is risk-reduction
Airlines don't care about risk, they never have. They care about cost.
Which is understandable, they are a business after all.
Risk and safety directly affect the cost outcome though.
Join Date: Feb 2015
Location: Alternate places
Age: 76
Posts: 97
Likes: 0
Received 0 Likes
on
0 Posts
Cost reduction the primary reason?
For most legacy carriers, the payroll alone for flight crews is second only to fuel costs, (debatable today - with both having plummetted over the differing periods of time, it's perhaps a toss-up which is lower, but they're not third or fourth). The industry is finding the pipelines drying up and autonomous flight provides an enticement not previously, realistically available.
As I have posted in earlier contributions, this is something like aviation's "Turing test"; there are obvious impediments to the notion of autonomous flight, the first being the thinking that someone on the ground writing software and firmware is somehow equal-to/better-than someone in the cockpit who is highly-trained, experienced, human and there, (with skin in the game).
These impediments are, in my (untrained/inexperienced in AI) eye, presently insurmountable but if one accepts that the fatal accident rate may climb "acceptably", (this is a perception/insurance/social issue, not a technical issue), then trials and targeted implementation may not be insurmountable while the concept is established towards "normal".
Some may accept that self-parking cars, UAVs and the like are "equivalents" to these goals; I think such perceptions are just magical thinking, not that dreaming is bad, but in aviation, there are no such things as sky-hooks, even in imaginative solutions to the problems of flight.
Airlines don't care about risk,
Join Date: Feb 2015
Location: Alternate places
Age: 76
Posts: 97
Likes: 0
Received 0 Likes
on
0 Posts
"Really?"
Yes.
I was incorrect only in stating "payroll for flight crews" - air carriers' highest financial liabilities are fuel, followed by employee (not just flight crew), salaries/benefits/costs. Flight crews are (or used to be) a substantial portion of the category. Fuel costs are about 5 to 10% higher than employee costs.
This does not take into account training costs & other costs associated with maintaining crews in cockpits.
Take a look at the financial statements of air carriers; many are available online. Airports/costs are way down the list.
Yes.
I was incorrect only in stating "payroll for flight crews" - air carriers' highest financial liabilities are fuel, followed by employee (not just flight crew), salaries/benefits/costs. Flight crews are (or used to be) a substantial portion of the category. Fuel costs are about 5 to 10% higher than employee costs.
This does not take into account training costs & other costs associated with maintaining crews in cockpits.
Take a look at the financial statements of air carriers; many are available online. Airports/costs are way down the list.
"Really?"
Yes.
I was incorrect only in stating "payroll for flight crews" - air carriers' highest financial liabilities are fuel, followed by employee (not just flight crew), salaries/benefits/costs. Flight crews are (or used to be) a substantial portion of the category. Fuel costs are about 5 to 10% higher than employee costs.
This does not take into account training costs & other costs associated with maintaining crews in cockpits.
Take a look at the financial statements of air carriers; many are available online. Airports/costs are way down the list.
Yes.
I was incorrect only in stating "payroll for flight crews" - air carriers' highest financial liabilities are fuel, followed by employee (not just flight crew), salaries/benefits/costs. Flight crews are (or used to be) a substantial portion of the category. Fuel costs are about 5 to 10% higher than employee costs.
This does not take into account training costs & other costs associated with maintaining crews in cockpits.
Take a look at the financial statements of air carriers; many are available online. Airports/costs are way down the list.
It's a big difference LOCO vs Legacy. But also hard to make direct comparison as does BA Employee costs cover the whole operation, or just flight & cabin crew. Wouldn't say airport costs are way down the list though, very near the top of both.
EZY Operating Costs
Fuel - £1,199 million
Airports and ground handling - £1,122 million
Crew - £505 million
Maintenance - £229 million
BA Operating Costs
Fuel, oil costs - £3,515 million
Employee costs - £2,422 million
Landing & handling costs - £2,168 million
- Landing fees and en route charges - £787 million
- Handling charges, catering and other operating costs - £1,381 million
Engineering and other aircraft costs - £613 million
Starting to shift rather a lot off the topic though.
Join Date: Feb 2015
Location: Alternate places
Age: 76
Posts: 97
Likes: 0
Received 0 Likes
on
0 Posts
LlamaFarmer;
Ah, ok - yes, LoCo, and U-LoCo, understand. BA's numbers look similar to here, except the landing fees & enroute costs, a billion+ USD? And of course, those costs can't be automated and like fuel are at the behest of the provider.
Off-topic - thought of that too, but removed the comment as it's orthogonal to the cheaper-than-pilots thesis.
Interesting tho', ...perhaps I'm not reading closely enough, but in the industry literature, (AW&ST, AI, BCA, etc., magazines, not the academic work), I'm not reading or seeing even little filler blurbs regarding for example, just the exploration of autonomous commercial flight; the subject's just not 'present'.
Ah, ok - yes, LoCo, and U-LoCo, understand. BA's numbers look similar to here, except the landing fees & enroute costs, a billion+ USD? And of course, those costs can't be automated and like fuel are at the behest of the provider.
Off-topic - thought of that too, but removed the comment as it's orthogonal to the cheaper-than-pilots thesis.
Interesting tho', ...perhaps I'm not reading closely enough, but in the industry literature, (AW&ST, AI, BCA, etc., magazines, not the academic work), I'm not reading or seeing even little filler blurbs regarding for example, just the exploration of autonomous commercial flight; the subject's just not 'present'.
Last edited by FDMII; 15th Jan 2016 at 01:55.
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
Uplinker
We have had this debate before.
On the previous thread you ignored all the links proving that all of your points were totally incorrect and provided no references of your own to back up any of your points.
I see little point in reliving that ignorance demo.
Suffice to say.
All your points are still wrong, and feel free to go to the previous discussion and this time read the links, watch the videos.
p.s. Seriously? Arguing that because a computer on the Apollo missions needed assistance that todays computers can't cope!?!?!
You do understand that things have moved on a little?
We have had this debate before.
On the previous thread you ignored all the links proving that all of your points were totally incorrect and provided no references of your own to back up any of your points.
I see little point in reliving that ignorance demo.
Suffice to say.
All your points are still wrong, and feel free to go to the previous discussion and this time read the links, watch the videos.
p.s. Seriously? Arguing that because a computer on the Apollo missions needed assistance that todays computers can't cope!?!?!
You do understand that things have moved on a little?
Oh dear. Got out the wrong side of the bed?
Since you quoted NASA - saying that their computer system would have totally coped with the Sioux City incident - I thought I would give you an example where a NASA computer got it very wrong. It did so because the wrong data - or insufficient data - had been loaded into it, (by humans). When it got to a few ?hundred feet above the moon, the pilots could see that the computer was about to do the wrong thing so they intervened.
The point I am trying to get across is that no matter how complicated and powerful a computer might be, it will still be built and programmed by humans, and the actions it takes will have been decided by humans. Humans are therefore still involved and humans make mistakes.
Your point of view seems to be based on an assumption that there wil never be any manufacturing or programmimg errors in a computer. I am challenging that point of view. You will always need humans to intervene if it all goes wrong.
Thank you for calling me ignorant, how nice. Do you know how computers work? Have you been trained in electronics,? Have you ever programmed a computer? (I can say yes to all three - and I am also an Airbus pilot approaching 10,000 hours TT, so I do have some experience in this area).
Since you quoted NASA - saying that their computer system would have totally coped with the Sioux City incident - I thought I would give you an example where a NASA computer got it very wrong. It did so because the wrong data - or insufficient data - had been loaded into it, (by humans). When it got to a few ?hundred feet above the moon, the pilots could see that the computer was about to do the wrong thing so they intervened.
The point I am trying to get across is that no matter how complicated and powerful a computer might be, it will still be built and programmed by humans, and the actions it takes will have been decided by humans. Humans are therefore still involved and humans make mistakes.
Your point of view seems to be based on an assumption that there wil never be any manufacturing or programmimg errors in a computer. I am challenging that point of view. You will always need humans to intervene if it all goes wrong.
Thank you for calling me ignorant, how nice. Do you know how computers work? Have you been trained in electronics,? Have you ever programmed a computer? (I can say yes to all three - and I am also an Airbus pilot approaching 10,000 hours TT, so I do have some experience in this area).
Last edited by Uplinker; 15th Jan 2016 at 09:53.
Join Date: Jun 2000
Location: last time I looked I was still here.
Posts: 4,507
Likes: 0
Received 0 Likes
on
0 Posts
APNewsBreak: Government not ensuring pilot skills are sharp
Not sure if this has been posted on this topic, but the topic on Rumours was very inactive. There are a couple of telling comments made by FAA and ideas to enforce manual skill enhancement including line flying. I know of a couple of large airlines who actively discourage such heresy. One reason is that too many screw it up and cause costs;y G/A's. Better to forbid than improve. The future is bleak.
Not sure if this has been posted on this topic, but the topic on Rumours was very inactive. There are a couple of telling comments made by FAA and ideas to enforce manual skill enhancement including line flying. I know of a couple of large airlines who actively discourage such heresy. One reason is that too many screw it up and cause costs;y G/A's. Better to forbid than improve. The future is bleak.
Join Date: Nov 2005
Location: Zulu Time Zone
Posts: 730
Likes: 0
Received 0 Likes
on
0 Posts
Uplinker
Is that really a good example? Computer with less capacity than a modern toaster has snag during most ground breaking test flight in history shocker!
Since you quoted NASA - saying that their computer system would have totally coped with the Sioux City incident - I thought I would give you an example where a NASA computer got it very wrong. It did so because the wrong data - or insufficient data - had been loaded into it, (by humans). When it got to a few ?hundred feet above the moon, the pilots could see that the computer was about to do the wrong thing so they intervened.
The Apollo 11 point is a red herring. The computer was not designed to perform a fully automatic landing without astronaut input. It was designed to provide FBW stabilisation, thrust control and guidance until the commander could see an intended landing site, at which point it was flown in a (still FBW) attitude hold and autothrust program to allow the commander to fly the vehicle to his chosen landing site.
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
There is no point guys.
You can tell him till you are blue in the face. You can provide references and videos but he will still come on the next thread an disagree as if the point was not proved.
The NASA example and the 10000 Airbus reference just make my point.
(You think 1970s vintage avionics in an airbus are somehow relevant?!)
You can tell him till you are blue in the face. You can provide references and videos but he will still come on the next thread an disagree as if the point was not proved.
The NASA example and the 10000 Airbus reference just make my point.
(You think 1970s vintage avionics in an airbus are somehow relevant?!)
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
Plus of course even a cursory knowledge of the Apollo landing would show that the pilot (Aldrin?) nearly screwed it up by being human. He messed with the computer when he should not have according to SOP and caused it buffer overrun failures.