The Naked Scientists

The Naked Scientists Forum

Author Topic: Can the procedure for access to flight decks be improved?  (Read 4733 times)

Offline syhprum

  • Neilep Level Member
  • ******
  • Posts: 3822
  • Thanked: 19 times
    • View Profile
With the two recent cases of two apparently mentally disturbed pilots deliberately crashing passenger aircraft it seems that the measures to prevent access to flight decks by terrorists are counter productive suggest improvements.
« Last Edit: 26/03/2015 22:06:48 by chris »


 

Offline chris

  • Neilep Level Member
  • ******
  • Posts: 5339
  • Thanked: 65 times
  • The Naked Scientist
    • View Profile
    • The Naked Scientists
Re: Can the procedure for access to flight decks be improved?
« Reply #1 on: 26/03/2015 22:07:59 »
We were debating this at work today. One suggestion might be to have a code system that, combined with sequences from the ground, could be used to open doors / activate systems or even take control of aeroplanes remotely. One has to balance the risk / benefit of such a situation.
 

Offline alancalverd

  • Global Moderator
  • Neilep Level Member
  • *****
  • Posts: 4715
  • Thanked: 154 times
  • life is too short to drink instant coffee
    • View Profile
Re: Can the procedure for access to flight decks be improved?
« Reply #2 on: 27/03/2015 00:00:13 »
For every terrorist or depressed pilot on an airliner, there are 200 bored blokes who just want to get to work, quickly and quietly. Forget cockpit doors and "security" screening: issue every passenger with a knife.
 

Offline Expectant_Philosopher

  • Sr. Member
  • ****
  • Posts: 229
    • View Profile
Re: Can the procedure for access to flight decks be improved?
« Reply #3 on: 27/03/2015 05:10:59 »
Could we improve automation of an aircraft for failsafe's to be built into the autopilot so that if the AI recognizes a crash parameter it would automatically seize control of the planes wing surfaces, to put the plane into a normal flight envelope?  If such a system had been in place for the Germanwings 9525 incident, the AI would've recognized a crash condition, taking into account the planes speed, GPS position, terrain map, and direction that the course of action initiated was not survivable, and would have altered wing control surfaces to recover the plane to a safe flight envelope.  I further think our black boxes are not enough measures to record an aircraft's flight.  With cloud computing and improved sensors, I believe even more data could be recovered from a flight, if during the flight the flight data were sent in bursts over a communications link to the ground for temporary storage until the aircraft's next flight.  I also think an aircraft should have a jettisonable beacon/drone that would provide the last known position of the aircraft and have on board the entire set of flight data, including voice and sensor readings.  And if a drone then the drone should attempt to track and follow the aircraft after jettison. 
 

Offline alancalverd

  • Global Moderator
  • Neilep Level Member
  • *****
  • Posts: 4715
  • Thanked: 154 times
  • life is too short to drink instant coffee
    • View Profile
Re: Can the procedure for access to flight decks be improved?
« Reply #4 on: 27/03/2015 07:51:11 »
Little difference between a "crash parameter" and a normal or emergency landing configuration, except for the presence of a mountain or the bit of the Hudson you have chosen to ditch in. Even with terrain recognition, the difference between a crash into a hill and an emergency uphill landing (or even a normal landing on a sloping runway) is a matter of seconds: a late override will break the plane anyway. And the problem with permanently-engaged  autopilots is that (1) they are quite happy to stall the aircraft in severe turbulence, when a human pilot would (should) simply accept inevitable departures from the assigned height (I think this was the Air Asia problem)  and (2) long periods of envelope protection can lead to complacency and loss of situational awareness when things go slightly adrift (Air France).

Looking at the Germanwings wreckage I dobt that anything could have been "jettisoned". Longhaul jets do transmit data bursts but most pilot suicides involve rather smaller aircraft, and there's not a lot you can do with the data from 2000 miles away unless you want all airliners to be completely remote-piloted. Which is an absolute gift for any terrorist: imagine being able to crash 200 planes at once!

The advantage of having a bloke at the front of the plane is that 99.999% of the time his principal interest is to get there and back safely, so everyone else gets a good ride. This leaves just two bad apples: the malevolent and the depressed. Not much you can do about malevolence, but I'd like to keep some time-expired trainers in a desert so that anyone who wants to end his life doing what he likes best, can do so without taking farepaying passengers with him.   

PS the one time I was persuaded to let a very simple autopilot handle the initial climbout, it tried to barrel-roll the aircraft as soon as we left the ground. So we disengaged, landed, and tried again: same problem. A limited autoland system is a great stress reliever, but machines do not generally recognise their own arrogance, and it is important that the pilot can override the gadget - otherwise you wouldn't have the benefit of my delightful prose!   
 

Offline evan_au

  • Neilep Level Member
  • ******
  • Posts: 4123
  • Thanked: 245 times
    • View Profile
Re: Can the procedure for access to flight decks be improved?
« Reply #5 on: 27/03/2015 10:26:08 »
Quote from: Expectant_Philosopher
I further think our black boxes are not enough measures to record an aircraft's flight.
Europe is a very closely monitored airspace, so there was no way that this particular plane would get "lost". In this case, the cockpit voice recorder (one of two black boxes) was rapidly recovered and the data analysed.

This is very different from Flight 370, which disappeared into the Southern Indian Ocean, far from air traffic control radars. At this point in time, nobody is publicly admitting that they know where MH370 now lies. But once the location is identified, the black box will provide considerable information in tracing the flight.

Over the years, the black box has been invaluable in investigating accidents - to the extent that some insurance companies are now providing an "app" that can be loaded on a car driver's smartphone. This acts as a black box, using the GPS for location, and accelerometers to detect impacts (conceivably, the phone's microphone and camera could provide additional information, but that may really cause concerns about "Big Brother"!). Some insurance companies are aiming to be notified of an accident within 2 minutes - so they call you before you call them!

Quote
if during the flight the flight data were sent in bursts over a communications link to the ground
I think that many players in the industry are now thinking that this is a good idea, and it is getting much easier with high-capacity satellite links with almost-global coverage.

But if the satellite link is still under pilot control, there is a risk it could be turned off by the pilot (just as Flight 370 turned off its radar transponder).

Quote
if the AI recognizes a crash parameter...

Most modern airliners have a series of warnings, covering things like "you are approaching the ground, and you haven't put your wheels down, yet". So even if the autopilot is not in control, the plane will warn the pilot of potentially dangerous situations.

But with today's technology, the pilot is still needed to adapt to unexpected situations.
 

Offline David Cooper

  • Neilep Level Member
  • ******
  • Posts: 1505
    • View Profile
Re: Can the procedure for access to flight decks be improved?
« Reply #6 on: 27/03/2015 17:00:43 »
If a plane is doing something it clearly shouldn't, it ought to be possible to override the pilot remotely and have the plane fly itself home. All members of the flight crew would have some code they can type in somewhere to enable the plane to take over control, and any two of them could make it happen, though for additional safety reasons it could also depend on a signal being sent to air traffic control so that they can check that the plane is actually doing something crazy at the time and talk to the pilots before giving the plane permission to override them. This system would not be of any use to terrorists who want to take over the plane.
 

Offline alancalverd

  • Global Moderator
  • Neilep Level Member
  • *****
  • Posts: 4715
  • Thanked: 154 times
  • life is too short to drink instant coffee
    • View Profile
Re: Can the procedure for access to flight decks be improved?
« Reply #7 on: 28/03/2015 00:00:14 »
If a plane is doing something it clearly shouldn't, it ought to be possible to override the pilot remotely and have the plane fly itself home.

Depends on your definition of "clearly shouldn't".

Loss of cabin pressure: descend immediately. Now your system takes over, returns the plane to cruising altitude and kills everyone.

Active thundercloud ahead: sharp right turn. System restores original flight path and kills everyone.

Converging aircraft: ditto

Sick passenger: divert to nearest medical facility. OK, only one death and everyone else arrives on time.

Engine failure: reduce cruise speed. System takes over, applies full power to remaining engine, pumps fuel into dead engine, and plane flies round in circles with dead engine ablaze until everyone dies.

Arrive early (tailwind), go into holding pattern. System takes over and flies plane direct to destination, colliding with traffic on scheduled approach and everyone dies.

Intercepted by military (not necessarily a bad thing - you may have signalled nav/com failure or hijack and requested an escort): turn to follow interceptor. System takes over, restores original heading. Military wingman releases missile and everyone dies.

Warning of airprox. Both aircraft turn sharp right to avoid,. Two systems take over and restore original headings, and everyone dies.
« Last Edit: 28/03/2015 09:23:34 by alancalverd »
 

Offline Bored chemist

  • Neilep Level Member
  • ******
  • Posts: 8667
  • Thanked: 42 times
    • View Profile
Re: Can the procedure for access to flight decks be improved?
« Reply #8 on: 28/03/2015 14:35:33 »
If a plane is doing something it clearly shouldn't, it ought to be possible to override the pilot remotely and have the plane fly itself home.

Depends on your definition of "clearly shouldn't".

Loss of cabin pressure: descend immediately. Now your system takes over, returns the plane to cruising altitude and kills everyone.

Active thundercloud ahead: sharp right turn. System restores original flight path and kills everyone.

Converging aircraft: ditto

Sick passenger: divert to nearest medical facility. OK, only one death and everyone else arrives on time.

Engine failure: reduce cruise speed. System takes over, applies full power to remaining engine, pumps fuel into dead engine, and plane flies round in circles with dead engine ablaze until everyone dies.

Arrive early (tailwind), go into holding pattern. System takes over and flies plane direct to destination, colliding with traffic on scheduled approach and everyone dies.

Intercepted by military (not necessarily a bad thing - you may have signalled nav/com failure or hijack and requested an escort): turn to follow interceptor. System takes over, restores original heading. Military wingman releases missile and everyone dies.

Warning of airprox. Both aircraft turn sharp right to avoid,. Two systems take over and restore original headings, and everyone dies.

You missed "ground avoidance radar goes wrong and says you are going to crash. Automatic system pulls up nose, stalls plane and everybody dies.
ditto for quite a lot of possible "sensor" failures.

How about giving all the passengers a console to fly the plane collectively?
 

Offline David Cooper

  • Neilep Level Member
  • ******
  • Posts: 1505
    • View Profile
Re: Can the procedure for access to flight decks be improved?
« Reply #9 on: 28/03/2015 17:39:20 »
If a plane is doing something it clearly shouldn't, it ought to be possible to override the pilot remotely and have the plane fly itself home.

Depends on your definition of "clearly shouldn't".

If one of the pilots has been shut out of the cockpit and air traffic control can see it decending into mountains, there is no danger of misinterpretation - it is people who would decide to tell the plane to override the person at the controls and to fly itself, and that's a better option in any case where the person at the controls intends to crash it into the ground at full speed on purpose.

Quote
Loss of cabin pressure: descend immediately. Now your system takes over, returns the plane to cruising altitude and kills everyone.

You're talking about a computer taking over without being asked to by a human, so it's a different case and irrelevant to what I said in my post above, but clearly if you wanted an intelligent computer to take over the machine you would require it to have sufficient wit to realise that the cabin has depressurised and to act accordingly by descending to a safe, lower altitude (assuming there is enough fuel left to reach a runway given that more fuel will be burnt flying at lower altitude).

Quote
Active thundercloud ahead: sharp right turn. System restores original flight path and kills everyone.

Again irrelevant for the same reason, but if you want a computer to override a pilot for going off the route he's suppost to be on, you would make sure the intelligent computer understands the need to avoid all possible hazards, including storms.

The same applies to all the rest of your suggestions - you are attacking the idea of putting a stupid computer in charge of everything and that is something which no one is suggesting should be done, yet you try to make it look as if it's something I'm proposing. You keep doing this kind of thing when replying to my posts and it's getting very irritating.
 

Offline alancalverd

  • Global Moderator
  • Neilep Level Member
  • *****
  • Posts: 4715
  • Thanked: 154 times
  • life is too short to drink instant coffee
    • View Profile
Re: Can the procedure for access to flight decks be improved?
« Reply #10 on: 28/03/2015 22:00:09 »


If one of the pilots has been shut out of the cockpit
Quote
we only knew this after the crash. ATC does not monitor cockpit conversations
and air traffic control can see it decending into mountains,
Quote
or an emergency airport, or just descending to a safe pressure altitude, in a valley between the mountains...




You're talking about a computer taking over without being asked to by a human,
Quote
so humans on the ground (which one? I talk to 3 different ATCs in a 100 mile journey, and there is very little ATC contact on a transpacific flight) are less inclined to do stupid things than pilots in the air?
so it's a different case and irrelevant to what I said in my post above, but clearly if you wanted an intelligent computer to take over the machine you would require it to have sufficient wit to realise that the cabin has depressurised and to act accordingly by descending to a safe, lower altitude (assuming there is enough fuel left to reach a runway given that more fuel will be burnt flying at lower altitude).
Quote
irrelevant assumption. It's better to land in a field, or even ditch in the sea, than asphyxiate


Again irrelevant for the same reason, but if you want a computer to override a pilot for going off the route he's suppost to be on, you would make sure the intelligent computer understands the need to avoid all possible hazards, including storms.
Quote
If only we could, we wouldn't need a pilot at all. Not all storms are hazardous or avoidable.

The same applies to all the rest of your suggestions - you are attacking the idea of putting a stupid computer in charge of everything
Quote
absolutely. There is no other sort of computer.
and that is something which no one is suggesting should be done, yet you try to make it look as if it's something I'm proposing.
Quote
nothing ad hominem, I assure you. Just examining the consequences of a radical change to what, until recently, has been a very satisfactory system
You keep doing this kind of thing when replying to my posts and it's getting very irritating
Quote
by no means as irritating as having a machine - or a man in an office - interrupt your lifesaving manouever and kill the passengers
.

Fact is that we can indeed fly ballistic missiles and passenger-carrying drones, and in the case of orbital rockets, we do so. But for some obscure reason (like hard experience) we prefer to have a pilot do things like landing and docking. Pilots generally put selfpreservation above all else, and it works quite well. Computers tend to put the supposed primary mission above any other consideration. "Get-home-itis" is a killer, whether it infects a man or a machine. So you need to devise a machine that can fly a plane and make sensible deviations from the primary mission as required: after 100 years of flight and 70 years of computing, the balance seems to favour the human. I've had satnavs, radio navs, autopilots and simple directional gyros fail in flight, but so far with no fatalities and not a lot of aggravation. Not sure how a computer would cope with conflicting inputs.   

Interestingly, the Germanwings kamikaze is alleged to have told his girlfriend that he wanted to be remembered. That's a bad sign. The best airline pilots want every flight to be completely unmemorable. Must go - I hope to sleep all the way to New York tomorrow (no, I'm not driving).
« Last Edit: 28/03/2015 22:24:47 by alancalverd »
 

Offline David Cooper

  • Neilep Level Member
  • ******
  • Posts: 1505
    • View Profile
Re: Can the procedure for access to flight decks be improved?
« Reply #11 on: 29/03/2015 20:04:15 »
If one of the pilots has been shut out of the cockpit

we only knew this after the crash. ATC does not monitor cockpit conversations

The pilot and flight crew knew before the crash. Which part of this is so hard for you to understand? They knew, so they could have acted straight away had there been a mechanism in place to allow them to. They had no way of overriding the copilot once he had locked the pilot out. My point was that there should be a mechanism by which the pilot (when shut out of the cockpit) and the rest of the flight crew should be able to switch out the copilot and get the plane to fly itself home. Despite the change in the rules, the same kind of incident could easily happen again - a fight between the pilots could lead to a rogue pilot killing the other in a fight while no one can get in from outside to intervene. There will still be no mechanism in place to prevent that because of stupidity with people again thinking they've solved the problem when they haven't.

Quote
Quote
and air traffic control can see it decending into mountains,

or an emergency airport, or just descending to a safe pressure altitude, in a valley between the mountains...


If the pilot, who has been locked out of the cockpit, describes the situation to air traffic control (this would need a second radio link as he wouldn't be doing this from the cockpit), they can check to see whether the path the plane is taking is a sensible one. The purpose of putting them into the loop is to deal with a situation where a rogue pilot has been shut out of the cockpit and is trying to override a pilot who is doing the right thing - air traffic control would be in touch with both of them and could make their judgement based on whether the person flying the plane appears to be doing something sensible or crazy. As soon as the pilot shows signs of doing something that doesn't look sensible, the plane could be remotely triggered into taking over control of the flight.

Quote
Quote
You're talking about a computer taking over without being asked to by a human,

so humans on the ground (which one? I talk to 3 different ATCs in a 100 mile journey, and there is very little ATC contact on a transpacific flight) are less inclined to do stupid things than pilots in the air?

All they have to do is look to see if the plane appears to be being flown by the safe pilot or the one who might be trying to crash on purpose. That would not have been hard in this case because they could see at the time that this plane was doing something crazy and their attempted communications with the rogue pilot were ignored. (I'm basing that bit on what was said on the news, but it may be subject to revision as it has come through the filter of journalism which can distort the facts enormously.)

Quote
Quote
so it's a different case and irrelevant to what I said in my post above, but clearly if you wanted an intelligent computer to take over the machine you would require it to have sufficient wit to realise that the cabin has depressurised and to act accordingly by descending to a safe, lower altitude (assuming there is enough fuel left to reach a runway given that more fuel will be burnt flying at lower altitude).
irrelevant assumption. It's better to land in a field, or even ditch in the sea, than asphyxiate

There is a compromise between keeping at a height that will make it home with the passengers getting enough oxygen to survive without being fully conscious and flying at a height which will make them feel good until the plane is forced to crash a long way short of a runway. I had assumed you would have sufficient intelligence to work that out for yourself without having to have it spelt out to you.

Quote
Quote
Again irrelevant for the same reason, but if you want a computer to override a pilot for going off the route he's suppost to be on, you would make sure the intelligent computer understands the need to avoid all possible hazards, including storms.
If only we could, we wouldn't need a pilot at all. Not all storms are hazardous or avoidable.

If a pilot can tell the difference, software will soon be able to do so as well. It may not be able to yet, but I have not said at any point that we should be getting rid of pilots yet.

Quote
Quote
The same applies to all the rest of your suggestions - you are attacking the idea of putting a stupid computer in charge of everything
absolutely. There is no other sort of computer.

The hardware's fine. Software is as stupid as the person who writes it, but you keep attacking this on the basis that all manner of obvious things wouldn't be taken into account at all when in reality they would all be thought about in great detail even by a team of very poor programmers.

Quote
Quote
and that is something which no one is suggesting should be done, yet you try to make it look as if it's something I'm proposing.
nothing ad hominem, I assure you. Just examining the consequences of a radical change to what, until recently, has been a very satisfactory system

What you're doing is attacking my proposal by distorting it into a different argument altogether. I proposed a mechanism for taking control of a plane away from a rogue pilot so that it can fly itself home - this would be triggered by the rest of the flight crew in combination with air traffic control, and it wouldn't be of any help to terrorists who want to take over the plane. The one thing it can't do is stop a rogue pilot putting the plane into a spin right at the start from which it can't recover, so there will still be a need to focus on developing software to handle emergency manoeuvres to keep pilots within the capabilities of the plane. Even then, it may be impossible for the shut-out-pilot to fight the extreme forces to get to a position where he can activate any override, so again more intelligent systems will be required so that rogue pilots can be overridden without requiring a human to make the decisions. We maybe can't do that yet, but it's something that should be worked on.

Quote
Quote
You keep doing this kind of thing when replying to my posts and it's getting very irritating
by no means as irritating as having a machine - or a man in an office - interrupt your lifesaving manouever and kill the passengers

Which is totally irrelevant as it has nothing whatsoever to do with what I'm proposing.

Quote
Fact is that we can indeed fly ballistic missiles and passenger-carrying drones, and in the case of orbital rockets, we do so. But for some obscure reason (like hard experience) we prefer to have a pilot do things like landing and docking. Pilots generally put selfpreservation above all else, and it works quite well. Computers tend to put the supposed primary mission above any other consideration. "Get-home-itis" is a killer, whether it infects a man or a machine. So you need to devise a machine that can fly a plane and make sensible deviations from the primary mission as required: after 100 years of flight and 70 years of computing, the balance seems to favour the human.

In this specific case, the human was the rogue computer that crashed the plane. My proposal is aimed at making it more difficult for a rogue human computer to cause a crash by switching him out and letting the plane take over to fly itself home, which in most situations of this rare kind will be fully within the current capabilities of software.
 

The Naked Scientists Forum

Re: Can the procedure for access to flight decks be improved?
« Reply #11 on: 29/03/2015 20:04:15 »

 

SMF 2.0.10 | SMF © 2015, Simple Machines
SMFAds for Free Forums