"Pilotless airliners safer" - London Times article
Some of them are quite serious (such as dual FMC failures in the case of the 777), and could have been catastrophic if there hadn't been a flight crew there to take over and fly the aircraft manually.
Also, the FADEC and FBW computers are "Design Assurance Level (DAL) "A". That means flight critical - a significant s/w error is considered to be potentially catastrophic (e.g. unilaterally surging or shutting down the engines, or commanding a dive into the ground). Yes, we still find errors in the implementation, but to the best of my knowledge a DAL A system has never been identified as the cause of a crash.
The 777 FMC is not a DAL "A" system. It is designed and certified to a lower level - a level that assumes if it screws up the pilots can and will take corrective action (granted, not always a valid assumption).
Also, the 777 systems were designed over 20 years ago. Ancient in electronic terms (compare a 1992 cell phone to one of today's smart phones). In fact one of the larger challenges that Boeing faces today is sourcing those 20 year old electronic components to keep building airplanes.
As I said before, electronics are advancing at an exponential rate. Same thing with s/w development. FADEC s/w isn't people entering lines of code (hasn't been for a long time) - it's people laying out 'flow diagrams' of how the want the s/w to work. Highly specialized automated programs then turn those diagrams into code - which is then exhaustively tested to make sure it functions as intended.
As Tourist pointed out, a pilotless airplane doesn't necessarily need to be perfect, it just needs to better than an all too flawed human pilot.
When SLF are routinely riding to the airport in automated cars that have an accident/fatality rate orders of magnitude better what us flawed human drivers can accomplish today, just how many CFTs, or 777 crashing into a seawall because the human pilot couldn't be bothered to monitor airspeed, do you think they'd be willing to tolerate?
It won't happen this decade, it may not happen during my lifetime (and I expect another 40 years, minimum
). But it will happen.BTW, those of you noting the Star Ship Enterprise has human pilots, you do know that's science fiction
Nemo Me Impune Lacessit
Join Date: Jun 2004
Location: Derbyshire, England.
Posts: 4,091
Received 0 Likes
on
0 Posts
This subject has been done to death so many times now, here on PPRuNe.
It may come as a disappointment to some here but, for the foreseeable future, the pilotless airliner simply isn't going to happen and not only for all the computer capability reasons given but, equally, for the much more mundane reason of security.
Dedicated terrorist organisations will have no problem in raiding and taking over a ground control station, particularly in the more remote parts of the world but, with suicidal tendencies, they will be able to manage it in more civilised places too. Alternatively they simply have to produce a jammer that is more powerful than a remote controlling (relay?) station and disconnect a few aircraft from the controlling system, carnage will soon follow.
One final point, the cost of development and demonstration, to a point of safety where the insurance underwriters will come on board and insure pilotless aircraft, is possibly more expensive than keeping pilots on board. No insurance = no fly.
It may come as a disappointment to some here but, for the foreseeable future, the pilotless airliner simply isn't going to happen and not only for all the computer capability reasons given but, equally, for the much more mundane reason of security.
Dedicated terrorist organisations will have no problem in raiding and taking over a ground control station, particularly in the more remote parts of the world but, with suicidal tendencies, they will be able to manage it in more civilised places too. Alternatively they simply have to produce a jammer that is more powerful than a remote controlling (relay?) station and disconnect a few aircraft from the controlling system, carnage will soon follow.
One final point, the cost of development and demonstration, to a point of safety where the insurance underwriters will come on board and insure pilotless aircraft, is possibly more expensive than keeping pilots on board. No insurance = no fly.
Join Date: Mar 2014
Location: australia
Age: 81
Posts: 207
Likes: 0
Received 0 Likes
on
0 Posts
As a person who only had a PPL and is SLF it would seem that when the pitot tubes of the AirFrance plane blocked the computer would have crashed the plane by itself. It instead disconnected and gave the chance for an experienced pilot to recover the situation it could not handle. Sadly it seems there was not an experienced pilot at the controls.
On youtube there is a movie about round the world in 66 hrs. A kind of promo for Lufthansa.
There are at least two incidents where human intervention became needed to prevent possible serious consequences due to the computers being able to control the situation. One was slow rate of ascent (Overweight?) at a fairly crucial waypoint and the other was a disconnect because of a serious overspeed condition.
I for one would like someone up front, preferably one with real experience but at a pinch I may have to accept someone with lots and lots of simulated training and not just in pushing buttons.
On youtube there is a movie about round the world in 66 hrs. A kind of promo for Lufthansa.
There are at least two incidents where human intervention became needed to prevent possible serious consequences due to the computers being able to control the situation. One was slow rate of ascent (Overweight?) at a fairly crucial waypoint and the other was a disconnect because of a serious overspeed condition.
I for one would like someone up front, preferably one with real experience but at a pinch I may have to accept someone with lots and lots of simulated training and not just in pushing buttons.
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
Wheels up and others.
Please do me the courtesy of bothering to read my links.
No, drones do not need a pilot.
Some currently are remotely piloted and some are fully autonomous and some are in between, ie a mouse click pilot.
I am talking about autonomous airliners.
You talk about drone crashes in the USA as if they are somehow proof that autonomous airliners can't be done, yet the article makes clear that most if not all of the accidents are pilot error or mechanical failure.
Pilot error is irrelevant for obvious reasons, and since the current crop of reaper/predator etc are single engined they are hardly relevant and will obviously have a failure rate closer to a Cessna than a Boeing.
You state that automation is currently moving the other way with increased training of pilots.
Surely you realise this is to deal with a weakness of pilots, not automation! Humans are very poor at dealing with situations we have not regularly practised, such as emergencies or even manual thrust etc. the current state of automation is very difficult for humans. We are expected to let automation carry out nearly all the work yet deal with the situation if the automation finds the going tough. Automation that is decades old in capability. A couple of sims per year will never make us good. Barely competent at best.
You mention having to protect the ILS everywhere.
Why on earth do you think an autonomous airliner would need an ILS at all to land in IMC? The systems required to land without have existed for decades. ILS would be a nice backup to the GPS/INS/Doppler/IR/radar/ etc etc etc that the system would be continuously monitoring for its solution.
You mention TCAS
Why on earth do you think that a human could be better than the autopilot at following TCAS RAs?
As it is now, it tells you what to fly and you try to fly it. Exactly like flight director, and in exactly the same way it would be more accurate itself.
You ask who would bother to develop An autonomous airliner and I have already posted a link to a major manufacturer doing just that!
You mention weather avoidance and say it is near impossible yet if you had bothered to read the link....
http://www.baesystems.com/innovation...3D152hotkppk_4
Various people keep saying that the cost of the pilots is small.
Not true.
First let's start with the pay. Let's say 20 pilots per aircraft. Each of these needs to be paid, recruited, trained(requiring simulators)
That is the small fry though. The real cost is the huge chunk of the airframe required to house them and give them the info they need.
Remove the entire cockpit and all those heavy controls and screens and armoured doors etc and you get a couple of tons of extra freight every flight and a few rows of 1st class seats every flight. That adds up to a huge amount of cash.
Mentioning flights where automation failed and so did the pilots is hardly helping your case. As tdracer says, those systems are designed to hand control back to a pilot in the (incorrect!)expectation that they will be capable of sorting the problem.
Parabellum
You are talking about a remote controlled airliner.
That is not what people are trying to construct. Autonomous means just that.
Incidentally, you mention terrorists overrunning a control centre. In what way is that different from them overrunning ATC now and giving out bad instructions to cause a crash?
TCAS and EGPWS would be the current final defence against such an event now, and similar for an autonomous aircraft.
Harryw
It is not surprising that an aircraft designed to have humans involved requires human intervention.
The level of automation tech on board current airliners is Stone Age.
I hate to prick your bubble, but "lots and lots of simulator training" is not how I would describe the pitifully small amount of annual sim time given to pilots, and it is mostly just "pushing buttons"
Please do me the courtesy of bothering to read my links.
No, drones do not need a pilot.
Some currently are remotely piloted and some are fully autonomous and some are in between, ie a mouse click pilot.
I am talking about autonomous airliners.
You talk about drone crashes in the USA as if they are somehow proof that autonomous airliners can't be done, yet the article makes clear that most if not all of the accidents are pilot error or mechanical failure.
Pilot error is irrelevant for obvious reasons, and since the current crop of reaper/predator etc are single engined they are hardly relevant and will obviously have a failure rate closer to a Cessna than a Boeing.
You state that automation is currently moving the other way with increased training of pilots.
Surely you realise this is to deal with a weakness of pilots, not automation! Humans are very poor at dealing with situations we have not regularly practised, such as emergencies or even manual thrust etc. the current state of automation is very difficult for humans. We are expected to let automation carry out nearly all the work yet deal with the situation if the automation finds the going tough. Automation that is decades old in capability. A couple of sims per year will never make us good. Barely competent at best.
You mention having to protect the ILS everywhere.
Why on earth do you think an autonomous airliner would need an ILS at all to land in IMC? The systems required to land without have existed for decades. ILS would be a nice backup to the GPS/INS/Doppler/IR/radar/ etc etc etc that the system would be continuously monitoring for its solution.
You mention TCAS
Why on earth do you think that a human could be better than the autopilot at following TCAS RAs?
As it is now, it tells you what to fly and you try to fly it. Exactly like flight director, and in exactly the same way it would be more accurate itself.
You ask who would bother to develop An autonomous airliner and I have already posted a link to a major manufacturer doing just that!
You mention weather avoidance and say it is near impossible yet if you had bothered to read the link....
http://www.baesystems.com/innovation...3D152hotkppk_4
Various people keep saying that the cost of the pilots is small.
Not true.
First let's start with the pay. Let's say 20 pilots per aircraft. Each of these needs to be paid, recruited, trained(requiring simulators)
That is the small fry though. The real cost is the huge chunk of the airframe required to house them and give them the info they need.
Remove the entire cockpit and all those heavy controls and screens and armoured doors etc and you get a couple of tons of extra freight every flight and a few rows of 1st class seats every flight. That adds up to a huge amount of cash.
Mentioning flights where automation failed and so did the pilots is hardly helping your case. As tdracer says, those systems are designed to hand control back to a pilot in the (incorrect!)expectation that they will be capable of sorting the problem.
Parabellum
You are talking about a remote controlled airliner.
That is not what people are trying to construct. Autonomous means just that.
Incidentally, you mention terrorists overrunning a control centre. In what way is that different from them overrunning ATC now and giving out bad instructions to cause a crash?
TCAS and EGPWS would be the current final defence against such an event now, and similar for an autonomous aircraft.
Harryw
It is not surprising that an aircraft designed to have humans involved requires human intervention.
The level of automation tech on board current airliners is Stone Age.
I hate to prick your bubble, but "lots and lots of simulator training" is not how I would describe the pitifully small amount of annual sim time given to pilots, and it is mostly just "pushing buttons"
Last edited by Tourist; 3rd Dec 2014 at 07:44.
Avoid imitations
Join Date: Nov 2000
Location: Wandering the FIR and cyberspace often at highly unsociable times
Posts: 14,573
Received 422 Likes
on
222 Posts
Technical backup for the first fully autonomous passenger airliner:
"AreOut
10 If both engines fail over city, try to land in river if runway not available.
That software didn't seem too difficult to write!
Next!"
then you would have many planes trying to land in river(and eventually cause havoc because rivers are sometimes crowded with ships) when they could actually reach the airport
see, you fail at the first step
please revert to your tourism and leave programming and piloting to professionals
10 If both engines fail over city, try to land in river if runway not available.
That software didn't seem too difficult to write!
Next!"
then you would have many planes trying to land in river(and eventually cause havoc because rivers are sometimes crowded with ships) when they could actually reach the airport
see, you fail at the first step
please revert to your tourism and leave programming and piloting to professionals
Surely you realise this is to deal with a weakness of pilots, not automation! Humans are very poor at dealing with situations we have not regularly practised
Would a computer have jumped a few steps or just followed the "programme"?
Finds Capt. Sully Sullenberger Could Have Landed at LaGuardia Airport
Seen it happen; fortunately the now #2 had all turning BUT was low on fuel
Avoid imitations
Join Date: Nov 2000
Location: Wandering the FIR and cyberspace often at highly unsociable times
Posts: 14,573
Received 422 Likes
on
222 Posts
Incidentally, you mention terrorists overrunning a control centre. In what way is that different from them overrunning ATC now and giving out bad instructions to cause a crash?
Don't forget, there has to be human input to program the autonomous aircraft in the first place.
Join Date: Jan 2008
Location: UK
Age: 73
Posts: 92
Likes: 0
Received 0 Likes
on
0 Posts
I might just get on a flight driven by software that is at version 99.9 but would I get on version 100.0?
The software industry is incredibly bad at testing new releases - how many times have we seen one new or updated feature cause mahem to something else after an update.
What a lonely bloody job that would be!
Nemo Me Impune Lacessit
Join Date: Jun 2004
Location: Derbyshire, England.
Posts: 4,091
Received 0 Likes
on
0 Posts
Incidentally, you mention terrorists overrunning a control centre. In what way is that different from them overrunning ATC now and giving out bad instructions to cause a crash?
This thread has drifted into a couple of areas that I know something about so I'm going to stop sitting on my hands.
I don't know whether fully automated commercial airliners will be realised before the aviation fuel runs out, but if they do arrive under current regulations then I'd be quite sure of the following:
I don't know whether fully automated commercial airliners will be realised before the aviation fuel runs out, but if they do arrive under current regulations then I'd be quite sure of the following:
- They will be 10 to 100 times safer than human piloted craft, because nothing less would be allowed..
- They will use "brute force" logic rather than AI, because you couldn't build a safety case around AI.
- They will not mix with human piloted air traffic, because that introduces too many variables.
Join Date: Sep 1999
Location: Limoges/Sussex
Posts: 21
Likes: 0
Received 0 Likes
on
0 Posts
It IS coming
http://www.astraea.aero/downloads/AS...d%20_FINAL.pdf
Tucked away in the original article is a short reference to the Astraea program which, provided you can maintain an open mind, is definitely worth researching. The above link is one starting point.
It's clear that the end goal is fully pilot-less air transport and it is probable that this will come, albeit by degrees. We just don't know the timescale yet.
Tucked away in the original article is a short reference to the Astraea program which, provided you can maintain an open mind, is definitely worth researching. The above link is one starting point.
It's clear that the end goal is fully pilot-less air transport and it is probable that this will come, albeit by degrees. We just don't know the timescale yet.
The hurdle for unmanned airliners is not technology.
The hurdle is passenger perception.
The hurdle is passenger perception.
*Cargo is where I can see this being accepted more readily. Amazon can drop as much junk on their way to my house as they want. I'm only paying for the one drone that makes it.
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
EEngr
Yes, I quite agree as I said before.
After the cargo fleet has flown for lets say a decade and demonstrated a better safety record then that is when the push will come for passenger aircraft.
Yes, I quite agree as I said before.
After the cargo fleet has flown for lets say a decade and demonstrated a better safety record then that is when the push will come for passenger aircraft.
Avoid imitations
Join Date: Nov 2000
Location: Wandering the FIR and cyberspace often at highly unsociable times
Posts: 14,573
Received 422 Likes
on
222 Posts
They will not mix with human piloted air traffic, because that introduces too many variables.
Join Date: Aug 2009
Location: England
Posts: 98
Likes: 0
Received 0 Likes
on
0 Posts
Flew Emirates a few months ago. On one flight just before take off an announcement was made "Hi there, I'm captain xxxxxxx, and I'm from Los Angeles, etc etc.
Slept well on that flight, somehow that particular captain had a re-assuring voice !!!!!!
For me, no captain no fly, simple.
Slept well on that flight, somehow that particular captain had a re-assuring voice !!!!!!
For me, no captain no fly, simple.
Join Date: Sep 2013
Location: UK
Posts: 730
Likes: 0
Received 0 Likes
on
0 Posts
This quote from Ozzy is the biggest factor:
"The fact that pilots are considered flawed, responsible for (typically 70%) of fatal aircraft accidents, and thus unmanned aircraft would be safer, is an untenable conclusion that ignores the vast proportion of accident or incident-free flights achieved daily by the current manned commercial air transport industry, particularly in States with well-developed aviaiton safety practice. To those who quote the 70% figure I respond: "...but 100% of non-accident flights are the result of the pilots doing their job successfully". Non-accident flights utterly dwarf the proportion of accident flights."
The simple truth is that there are no official estimates, never mind reliable statistics, to demonstrate how many occasions human pilots have prevented automated disaster. My estimate is a couple of times per flight. Not only do they dwarf the pilot error events, they would make the pilot error accidents statistically irrelevant.
"The fact that pilots are considered flawed, responsible for (typically 70%) of fatal aircraft accidents, and thus unmanned aircraft would be safer, is an untenable conclusion that ignores the vast proportion of accident or incident-free flights achieved daily by the current manned commercial air transport industry, particularly in States with well-developed aviaiton safety practice. To those who quote the 70% figure I respond: "...but 100% of non-accident flights are the result of the pilots doing their job successfully". Non-accident flights utterly dwarf the proportion of accident flights."
The simple truth is that there are no official estimates, never mind reliable statistics, to demonstrate how many occasions human pilots have prevented automated disaster. My estimate is a couple of times per flight. Not only do they dwarf the pilot error events, they would make the pilot error accidents statistically irrelevant.