Opportunities, Challenges, and Limits of Automation in Aircraft
They have taken a lot of the day-to-day things and automated them and this has had the effect of weakening pilot skills and when it does drop its bundle you'll need those very same now weakened skills to fix it.
Join Date: Jun 2014
Location: Chocolatetown
Age: 63
Posts: 83
Likes: 0
Received 0 Likes
on
0 Posts
The Dynon Avionics Flightdek D180, Garmin GPS and Foreflight for iPad are sweet but are BACKUPS for my brain.
NNC Item # 1. Brain..................... Engage
Steam Gauges Are Safer - IFR Magazine Article
Interesting thread.
The problem with airline flying as it is now, is that when one starts line flying big jets at a new company or a new type, there is too much going on initially to practise live raw data flying on the line in busy airspace. And even though they might use old technology, current A/P, F/D and A/THR are very good, (certainly Airbus FBW), as long as one understands how the automatics work, and how to use them correctly
So one uses the F/Ds and the A/THR, and the A/P as well, while getting used to the new SOPs, routes, company culture etc, just to keep on top of it all.
Flying a modern jet by hand but using the F/D is a doddle, even FBW, (as long as the F/D has been set up correctly for the particular flight phase and it’s data is good). However, every day that one uses the F/D instead of raw data causes one’s raw data instrument scan and skills to atrophy a little more. After not very many sectors of using a F/D, one feels their raw data skills are perhaps a bit rusty and doesn’t want to embarrass themselves or worse, cause a go-around by attempting a raw data approach and screwing it up in front of a new colleague. So the F/D is used all the time.
So something needs to change to keep our raw data and hand flying skills alive. The SIM is an obvious place for this, and more raw data/hand flying should be done there. (And trained properly - I have never been trained to fly a big jet on raw data manually, just shouted at or tutted at if it goes wrong). But some incentive needs to be brought in to keep our skills sharp on the line in the real aircraft. I have proposed a system where we have to do a certain number of raw data hand flown approaches in a certain period - just like we used to do with Autolands. We would have to keep a personal record and have to show that we did 3 such approaches in the previous 6 months?
3 in 6 months would not be a lot of manual flying, but it would be a start, and might encourage pilots to fly manually more often.
The problem with airline flying as it is now, is that when one starts line flying big jets at a new company or a new type, there is too much going on initially to practise live raw data flying on the line in busy airspace. And even though they might use old technology, current A/P, F/D and A/THR are very good, (certainly Airbus FBW), as long as one understands how the automatics work, and how to use them correctly
So one uses the F/Ds and the A/THR, and the A/P as well, while getting used to the new SOPs, routes, company culture etc, just to keep on top of it all.
Flying a modern jet by hand but using the F/D is a doddle, even FBW, (as long as the F/D has been set up correctly for the particular flight phase and it’s data is good). However, every day that one uses the F/D instead of raw data causes one’s raw data instrument scan and skills to atrophy a little more. After not very many sectors of using a F/D, one feels their raw data skills are perhaps a bit rusty and doesn’t want to embarrass themselves or worse, cause a go-around by attempting a raw data approach and screwing it up in front of a new colleague. So the F/D is used all the time.
So something needs to change to keep our raw data and hand flying skills alive. The SIM is an obvious place for this, and more raw data/hand flying should be done there. (And trained properly - I have never been trained to fly a big jet on raw data manually, just shouted at or tutted at if it goes wrong). But some incentive needs to be brought in to keep our skills sharp on the line in the real aircraft. I have proposed a system where we have to do a certain number of raw data hand flown approaches in a certain period - just like we used to do with Autolands. We would have to keep a personal record and have to show that we did 3 such approaches in the previous 6 months?
3 in 6 months would not be a lot of manual flying, but it would be a start, and might encourage pilots to fly manually more often.
Last edited by Uplinker; 9th Dec 2018 at 10:15.
and might encourage pilots to fly manually more often.
Readers of US Flying magazine of years ago may remember Captain Len Morgan who wrote a column called "Vectors." He was a wonderful writer who flew 707 and 747 trips to Guam, Seoul, Tokyo, Vietnam, Hong Kong and other Far Eastern points.
I corresponded with Len before the internet arrived. We discussed the new automatics era which was just coming in with the introduction of the first EFIS Boeing 737's. Here is an extract of a letter I received from Len in 1995 in reply to concerns I had about degradation of pilot flying skills as automatics got ever more sophisticated.
Quote:
"Dear John,
Great minds think alike. I could not agree more with your comments on the new tendency to teach young pilots to rely on what you call the "automatics" instead their own flying skills. I saw this trend developing with the first 707's and spreading into the 727s and 747s. I am sure it has gone much further since I retired in 1982, in fact, I know it has since our son is a check airman with USAir on the 737-200. He deplores the trend and when checking out a new captain likes to turn off all the gadgets 100 miles out and tell the new man to find the field and land DC-3 fashion. No DME, no glide slope, no localizer, no VOR. Where the automatics get people into trouble is when a real problem develops. They start punching buttons instead of grabbing the wheel". Unquote.
There was more in his letter but you get my drift. While airline managements may pay lip service to the need to keep in practice at pure instrument flying skills without the aid of the automatics, in reality this rarely will be actively encouraged and for various reasons. Legalities for one. That leaves simulator practice and it is left to the pilot to insist on regular hands-on practice. Unfortunately, the politics of that will often interfere with the pilot's good intentions.
Centaurus, et al, for the objective of debate, I disagree about the need for more hand flying.
Much of the discussion looks to future wish-lists, based on existing operations (what is or what was - solve today’s problems), instead we need to look further forward and consider, ‘what could be’.
This would not be dominated by our (piloting) point of view, but those of society and the industry.
A major driver will be economics. Technology is already at a level where single pilot operations are viable, but this requires expensive proof of concept, certification, and social - public acceptance. I don’t see that latter as dominating the debate, thus the issue comes down to the cost of achieving change against the long term benefits; and an obstacle there is in our thinking.
Small cargo aircraft can be single pilot, but these operations are more likely to be challenged by fully automatic, autonomous, pilotless aircraft as the costs of procurement decrease. There are Nav, ATC, conflict problems to solve, but that’s the challenge.
Commuter operations may be single pilot plus a helper ‘pilot’ (no disrespect for helpers), thus changing to full automation plus single ‘helper’ is feasible with existing technologies, but cost of change has to be balanced; perhaps not just yet.
Large commercial aircraft could easily follow the ideas of commuter operations, but I judge that this might not be so readily accepted - certification, society; so there may be interim stage.
This would be the basis of a new commercial aircraft; two crew, fully automated, where pilot assistance is only required for those situations which have not been foreseen (which unfortunately appear in many recent accidents). The new pilot-systems operator (‘tolip’), would not need extensive physical flying skills, only those necessary to achieve an acceptable outcome for rare total systems failures - ‘if you can walk away from it’. Cognitive skills would be those associated with the new systems, not those based on current flying standards.
Thus the challenge is how to improve technological standards and levels of safety; the present is a pointer to the future.
A significant difficulty is in how much the industry assumes or expects from pilot intervention in exceptional circumstance. Currently many views would cite the pilot as the weak link, yet with thoughtful analysis, it is the human contribution elsewhere which falls down.
If we can understand these aspects, then, in time, the ‘new’ two crew aircraft could revert to one, and further to …… 0
There are arguments about incapacitation etc, but if automation improves along current lines then we pilots will fall over more frequently than the technology.
There will be problems in achieving these changes; where will the interim pilots come from before we reach the ‘tolip’ stage. Will the human in design be the weakest link, or in certification, manufacturing and maintenance. What will be the environmental views, demand for travel, alternative transport.
… the present is a pointer, ‘the key’, to the future… but that key is not more hand flying, it’s in the way we think about operating … safely.
Much of the discussion looks to future wish-lists, based on existing operations (what is or what was - solve today’s problems), instead we need to look further forward and consider, ‘what could be’.
This would not be dominated by our (piloting) point of view, but those of society and the industry.
A major driver will be economics. Technology is already at a level where single pilot operations are viable, but this requires expensive proof of concept, certification, and social - public acceptance. I don’t see that latter as dominating the debate, thus the issue comes down to the cost of achieving change against the long term benefits; and an obstacle there is in our thinking.
Small cargo aircraft can be single pilot, but these operations are more likely to be challenged by fully automatic, autonomous, pilotless aircraft as the costs of procurement decrease. There are Nav, ATC, conflict problems to solve, but that’s the challenge.
Commuter operations may be single pilot plus a helper ‘pilot’ (no disrespect for helpers), thus changing to full automation plus single ‘helper’ is feasible with existing technologies, but cost of change has to be balanced; perhaps not just yet.
Large commercial aircraft could easily follow the ideas of commuter operations, but I judge that this might not be so readily accepted - certification, society; so there may be interim stage.
This would be the basis of a new commercial aircraft; two crew, fully automated, where pilot assistance is only required for those situations which have not been foreseen (which unfortunately appear in many recent accidents). The new pilot-systems operator (‘tolip’), would not need extensive physical flying skills, only those necessary to achieve an acceptable outcome for rare total systems failures - ‘if you can walk away from it’. Cognitive skills would be those associated with the new systems, not those based on current flying standards.
Thus the challenge is how to improve technological standards and levels of safety; the present is a pointer to the future.
A significant difficulty is in how much the industry assumes or expects from pilot intervention in exceptional circumstance. Currently many views would cite the pilot as the weak link, yet with thoughtful analysis, it is the human contribution elsewhere which falls down.
If we can understand these aspects, then, in time, the ‘new’ two crew aircraft could revert to one, and further to …… 0
There are arguments about incapacitation etc, but if automation improves along current lines then we pilots will fall over more frequently than the technology.
There will be problems in achieving these changes; where will the interim pilots come from before we reach the ‘tolip’ stage. Will the human in design be the weakest link, or in certification, manufacturing and maintenance. What will be the environmental views, demand for travel, alternative transport.
… the present is a pointer, ‘the key’, to the future… but that key is not more hand flying, it’s in the way we think about operating … safely.
Last edited by safetypee; 9th Dec 2018 at 15:57.
I have a healthy distrust of automated systems
Join Date: Sep 2016
Location: USA
Posts: 803
Likes: 0
Received 0 Likes
on
0 Posts
This would be the basis of a new commercial aircraft; two crew, fully automated, where pilot assistance is only required for those situations which have not been foreseen (which unfortunately appear in many recent accidents). The new pilot-systems operator (‘tolip’), would not need extensive physical flying skills, only those necessary to achieve an acceptable outcome for rare total systems failures - ‘if you can walk away from it’. Cognitive skills would be those associated with the new systems, not those based on current flying standards.
Join Date: Jul 2009
Location: Not far from a big Lake
Age: 81
Posts: 1,454
Likes: 0
Received 0 Likes
on
0 Posts
The way to start getting pilot skills up to speed more is to fly more. The biggest discouragement to this process is RVSM airspace where Otto has to be in control by regulation.
Suppose we build in an "instructor mode" into the system where we can actually fly in RVSM airspace and the system monitors us closely enough to prevent excessive rates from developing by making control inputs once a tolerance band is on the way to being exceeded, ie. a predictive exceedance of the tolerance band. In essence a computer game for pilots to build skills while in the cockpit. You wouldn't have to fly the whole flight that way, just enough to keep your scan technique refreshed, perhaps only a few minutes at a time.
Any competitive pilot would try to keep the "instructor's" heavy hand off their controls by flying as carefully as possible. After a bit of this, type of flying, I suspect many pilots would only occasionally see correction from the instructor mode.
FWIW almost all my X-country jet time was hand flown. We didn't even maintain the autopilot in one variety of jet that I flew (single pilot)
Suppose we build in an "instructor mode" into the system where we can actually fly in RVSM airspace and the system monitors us closely enough to prevent excessive rates from developing by making control inputs once a tolerance band is on the way to being exceeded, ie. a predictive exceedance of the tolerance band. In essence a computer game for pilots to build skills while in the cockpit. You wouldn't have to fly the whole flight that way, just enough to keep your scan technique refreshed, perhaps only a few minutes at a time.
Any competitive pilot would try to keep the "instructor's" heavy hand off their controls by flying as carefully as possible. After a bit of this, type of flying, I suspect many pilots would only occasionally see correction from the instructor mode.
FWIW almost all my X-country jet time was hand flown. We didn't even maintain the autopilot in one variety of jet that I flew (single pilot)
Join Date: Nov 2018
Location: Planet Earth
Posts: 17
Likes: 0
Received 0 Likes
on
0 Posts
I expect that burden to be a huge issue in semiautomated cars, as long as it is allowed (as seemingly it currently is, hands always on wheel, but brain out of the window).
Join Date: Jun 2014
Location: Chocolatetown
Age: 63
Posts: 83
Likes: 0
Received 0 Likes
on
0 Posts
It will require expeditious recognition of erroneous output from incorrect/bad data. There isn't a checklist for that. You fly out of it manually.
Bad data can come from bad hardware (JT610), bad programming, bad inputs or any combination of these. The Sunwing Airlines Belfast incident is an example of bad data being input by the crew. OAT of -52c was input into the FMC. The FMC did what it was programmed to do. In JT610, a bad AoA sensor fed bad data (to a bad program I would argue) to the FMC. Again the FMC did what it was programmed to do.
NOT easily trained. Requires exceptional situational awareness.
Bad data can come from bad hardware (JT610), bad programming, bad inputs or any combination of these. The Sunwing Airlines Belfast incident is an example of bad data being input by the crew. OAT of -52c was input into the FMC. The FMC did what it was programmed to do. In JT610, a bad AoA sensor fed bad data (to a bad program I would argue) to the FMC. Again the FMC did what it was programmed to do.
NOT easily trained. Requires exceptional situational awareness.
Last edited by climber314; 9th Dec 2018 at 21:31.
Lots of pilots don't understand what EPR or N1 as a thrust setting parameter actually means beyond a textbook/ exam answer definition. N1 goes up in the climb, why? exactly. Why is EPR indicative of thrust, precisely? I've met people with 10000+ hours that have no idea. "it just does what i expect it to do" Yet these people still expect six figure salaries??
Last edited by Private jet; 10th Dec 2018 at 00:10.
I have a healthy distrust of automated systems. More so when my life enters the equation.
"The following story received today from a colleague of mine from years ago. It was so unusual and so important that I felt that readers would be interested
Quote:
"I had a major "learning" in Miami on takeoff as an observer when we got a stick shaker around 12 degrees N.U on take off. Just as we lined up I looked at the flight instruments and something wasn't right but I couldn'd identify it. Later in flight as all hell broke loose I noticed the left PLI/ "eyebrows" was significantly lower than the right side which appeared "normal".
This was the cause of the stick shaker (invalid AoA) and this must have been the case with the Lion Air 737 as well. Seeing this, might have saved that flight -putting aside the MCAS for a minute. Bottom line though is they should have not have attempted to fly with the AoA issue.
Back on the ground, I went directly to the mechanic who last worked on the 767 and he sheepishly admitted he installed a new AoA instrument the night before and had dropped it on the ground likely damaging it. It was stuck at a setting of 12 degrees. The normal setting is 15 degrees on the ground and it rises above this as you speed up/climb etc
This is knowledge every pilot should have. Too late for those poor 190 people on Lion Air."
Unquote.
Lots of pilots don't understand what EPR or N1 as a thrust setting parameter actually means beyond a textbook/ exam answer definition. N1 goes up in the climb, why? exactly. Why is EPR indicative of thrust, precisely? I've met people with 10000+ hours that have no idea. "it just does what i expect it to do" Yet these people still expect six figure salaries??
The “six figure salary” (I wish), is not for being an engine or avionics engineer, it is for being an experienced Captain and all that entails; with legal responsibility for the safety and management of the crew, passengers and aircraft under their charge. How much does a GP earn. - £200,000 ? Do they know and recognise every disease and malady without having to look it up? How much does a dentist or an MP or your local council leader earn?
In the past of course, airlines carried flying engineers who did know how all the systems worked. Then, economics decided to do away with them by computerising some systems and transferring everything to the overhead panel, to be controlled by the remaining two pilots. - whilst also flying the ‘plane. And other systems became much easier to use through developments such as FADECs.
Join Date: Jun 2014
Location: Chocolatetown
Age: 63
Posts: 83
Likes: 0
Received 0 Likes
on
0 Posts
Back on the ground, I went directly to the mechanic who last worked on the 767 and he sheepishly admitted he installed a new AoA instrument the night before and had dropped it on the ground likely damaging it. It was stuck at a setting of 12 degrees. The normal setting is 15 degrees on the ground and it rises above this as you speed up/climb etc
This is knowledge every pilot should have. Too late for those poor 190 people on Lion Air.
This is knowledge every pilot should have. Too late for those poor 190 people on Lion Air.
Does a 737 Max (without the AoA "option") even have this feature?
My understanding is there was no AoA Disagree message.
Seems like Boeing should make this "standard equipment" on the 737 in hindsight.
I digress.
Yes and no. Do you know the physics of, and how every part of your aircraft works? The FMC, the TRUs, the radios, the packs, the generators, the weather radar? Probably not. You probably know how to use the equipment but maybe not how it actually works.
The “six figure salary” (I wish), is not for being an engine or avionics engineer, it is for being an experienced Captain and all that entails; with legal responsibility for the safety and management of the crew, passengers and aircraft under their charge. How much does a GP earn. - £200,000 ? Do they know and recognise every disease and malady without having to look it up? How much does a dentist or an MP or your local council leader earn?
In the past of course, airlines carried flying engineers who did know how all the systems worked. Then, economics decided to do away with them by computerising some systems and transferring everything to the overhead panel, to be controlled by the remaining two pilots. - whilst also flying the ‘plane. And other systems became much easier to use through developments such as FADECs.
Last edited by Private jet; 12th Dec 2018 at 22:48.
AF447 was the result of icing causing a loss of air data, not automation, but it does have some relevance here. Pilot 'six figure salaries' and GP's bear none whatsoever.
447 was quite sad, never needed to happen and went well beyond normal pilot behaviour. There was never any need to go to those extreme attitudes or control positions and nothing about any of the control inputs that were made makes any sense. The fact that the side sticks did not give any feedback of the 'pilot's' inputs to the other pilot was also directly implicated.
BASIC aeronautical knowledge - Power+Attitude=Performance - every pilot is taught from the beginning, or at least was, that airspeed is a performance output which is the result of setting an attitude and a power, so, if you are at a NORMAL attitude with NORMAL power set, then performance will be NORMAL.
All they had to do was disconnect the automatics (AP, FD, A/T), set the power to where it had been for the past few hours (normal) and hold the attitude at 2.5 deg (normal) and also where it had been for the past few hours, and nothing would have happened. The air data would have eventually returned and then at that point they simply continued.
How could holding and sustaining full back stick at altitude for that length of time EVER RESULT IN ANYTHING OTHER than a deep stall?
Airbus seem to have tried to engineer pilots out of things as much as possible, and in so doing have made aircraft that are, in some situations, much more difficult to fly than they really need to be. Deliberately removing as much tactile feedback as possible (you don't need it, you can look at the gauges) and deliberately placing identical, dangerous, switches close to one another so that they look tidy or pretty is something engineers may like, but not pilots.
I think it was the chief designer of the A320 who was quoted at some point as saying that his aircraft are so easy and beautiful to fly that his local patisserie chef could fly one. Astonishingly, and perhaps quite irresponsibly, he seems happy with flying becoming a video game, and pilots becoming video gamers...... very, very wrong!
447 was quite sad, never needed to happen and went well beyond normal pilot behaviour. There was never any need to go to those extreme attitudes or control positions and nothing about any of the control inputs that were made makes any sense. The fact that the side sticks did not give any feedback of the 'pilot's' inputs to the other pilot was also directly implicated.
BASIC aeronautical knowledge - Power+Attitude=Performance - every pilot is taught from the beginning, or at least was, that airspeed is a performance output which is the result of setting an attitude and a power, so, if you are at a NORMAL attitude with NORMAL power set, then performance will be NORMAL.
All they had to do was disconnect the automatics (AP, FD, A/T), set the power to where it had been for the past few hours (normal) and hold the attitude at 2.5 deg (normal) and also where it had been for the past few hours, and nothing would have happened. The air data would have eventually returned and then at that point they simply continued.
How could holding and sustaining full back stick at altitude for that length of time EVER RESULT IN ANYTHING OTHER than a deep stall?
Airbus seem to have tried to engineer pilots out of things as much as possible, and in so doing have made aircraft that are, in some situations, much more difficult to fly than they really need to be. Deliberately removing as much tactile feedback as possible (you don't need it, you can look at the gauges) and deliberately placing identical, dangerous, switches close to one another so that they look tidy or pretty is something engineers may like, but not pilots.
I think it was the chief designer of the A320 who was quoted at some point as saying that his aircraft are so easy and beautiful to fly that his local patisserie chef could fly one. Astonishingly, and perhaps quite irresponsibly, he seems happy with flying becoming a video game, and pilots becoming video gamers...... very, very wrong!
Plastic PPRuNer
As I have said before, I am not a pilot but have done a lot of lab work involving programming.
Any program has to "sanitize" inputs before they enter the main body of the program.
If my subject's temperature has been a steady 37degC for the last two hours and suddenly starts reading 30degC then there is something wrong with the probe or it's connection or attachment to the subject.
And the program has to respond in a sensible way, i.e. NOT turn up the thermostat to 40degC in an attempt to warm the subject, but have been told that such an sudden drop is impossible and give me an appropriate informative warning - "UNEXPECTED TEMP FALL - CHECK PROBE ATTACHMENT AND CONNECTION"
Mac
Any program has to "sanitize" inputs before they enter the main body of the program.
If my subject's temperature has been a steady 37degC for the last two hours and suddenly starts reading 30degC then there is something wrong with the probe or it's connection or attachment to the subject.
And the program has to respond in a sensible way, i.e. NOT turn up the thermostat to 40degC in an attempt to warm the subject, but have been told that such an sudden drop is impossible and give me an appropriate informative warning - "UNEXPECTED TEMP FALL - CHECK PROBE ATTACHMENT AND CONNECTION"
Mac
Well, i did at one time hold an aircraft engineers licence, and now an ATPL. But things I gave an example of are BASIC aeronautical knowledge, and pilot's should know this. AF447, the only thing that agreed on the flightdeck were the 3 ADI's, why didn't they "fly attitude"? Piloting an aircraft in the modern era is not that difficult really, if you think it is maybe you should consider another profession. If the "responsibility" worries you the former applies too. By the way responsibility is only an issue if YOU make a mistake. So by inference, what you are saying is that you should be paid a lot of money so that you don't make a mistake? Kind of a bizarre reverse Mafia protection racket argument there. Why do you compare pilots to GP's? Apples and oranges surely? No comparison. GP's, in the UK at least, operate in a non competative market. Pilots do. Sorry for straying away from technical discussion, but some inane comments deserve to be replied to.
Your post to me is a straw man argument - I did not say the things you claim I did: you are putting words into my mouth and inferring what simply isn’t there. Where did I claim that “piloting an aircraft in the modern era is difficult”? (And thank you for kindly suggesting that I look for another profession, how thoughtful of you). Where did I say that responsibility worried me? And the inference you claim to make is frankly ridiculous and undeserving of a reply.
Oh, and adding a nice insult at the end. Lovely. Have another read, there’s a good chap.
This discussion of automation is also happening in the train and mining truck world, with some progress evident:
https://arstechnica.com/cars/2018/12...y-operational/
It would be logical to expect Amazons airfreight service to become the pioneer US testbed for autonomous commercial flight.
https://arstechnica.com/cars/2018/12...y-operational/
It would be logical to expect Amazons airfreight service to become the pioneer US testbed for autonomous commercial flight.
