Would planes be safer without humans in the cockpit?
Mar 27, 2015, 8:47 am
#31
Join Date: Aug 2006
Location: San Jose CA
Posts: 1,100
Cockpit Automation
The ever-witty pilot Patrick Smith has written about this topic extensively. Here's an article that explains why commercial aircraft aren't drones.
Last edited by boberonicus; Mar 27, 2015 at 2:12 pm
Mar 27, 2015, 9:37 am
#32
Join Date: Dec 2009
Location: RDU
Programs: DL DM+(segs)/MM, UA Ag, Hilton DM, Marriott Ti (life Pt), TSA Opt-out Platinum
Posts: 3,226
This is an interesting discussion.
On one hand, many past crashes (i.e. pilot error) could have likely been avoided with good automation (AF447, OH5191, Germanwings, OS214, etc).
On the other hand, it's impossible to program every possible unlikely scenario. UA232 comes to mind as well as US1547 that someone mentioned above and many other scenarios where a pilot was able to get a plane on the ground where the computer would not have understood the solution.
It would be interesting to go back and look at all the aircraft incidents in the last 20 or so years and see which could have been prevented by complete automation and which couldn't. I suspect the results would be surprising.
I do think we'll see pilotless A/C it in the next 20-30 years, but it will take a lot of time and investment and we'll still need pilots (even if controlling from the ground, like drones).
On one hand, many past crashes (i.e. pilot error) could have likely been avoided with good automation (AF447, OH5191, Germanwings, OS214, etc).
On the other hand, it's impossible to program every possible unlikely scenario. UA232 comes to mind as well as US1547 that someone mentioned above and many other scenarios where a pilot was able to get a plane on the ground where the computer would not have understood the solution.
It would be interesting to go back and look at all the aircraft incidents in the last 20 or so years and see which could have been prevented by complete automation and which couldn't. I suspect the results would be surprising.
I do think we'll see pilotless A/C it in the next 20-30 years, but it will take a lot of time and investment and we'll still need pilots (even if controlling from the ground, like drones).
Mar 27, 2015, 2:05 pm
#33
Join Date: Feb 2013
Location: Somewhere In The Five Eyes
Posts: 228
> If you notice that software updates are released fairly regularly for mobile
> phones, computers, general software ...
Ditto for airplanes and avionics.
> and sometimes that software update breaks things in an effort to fix things.
Ditto for airplanes and avionics.
> phones, computers, general software ...
Ditto for airplanes and avionics.
> and sometimes that software update breaks things in an effort to fix things.
Ditto for airplanes and avionics.
Mar 27, 2015, 2:20 pm
#34
FlyerTalk Evangelist
Join Date: May 2006
Location: SEA or BGR, Lower Earth Orbit
Programs: UA 1K
Posts: 17,217
> If you notice that software updates are released fairly regularly for mobile
> phones, computers, general software ...
Ditto for airplanes and avionics.
> and sometimes that software update breaks things in an effort to fix things.
Ditto for airplanes and avionics.
> phones, computers, general software ...
Ditto for airplanes and avionics.
> and sometimes that software update breaks things in an effort to fix things.
Ditto for airplanes and avionics.
Mar 27, 2015, 7:15 pm
#35
Original Member
Join Date: May 1998
Location: NYC
Programs: AA 2MM, Bonvoy LTT, Hilton Diamond
Posts: 14,636
Good luck selling that to the public.
Mar 28, 2015, 1:47 am
#36
Join Date: Jul 2007
Location: San Francisco/Sydney
Programs: UA 1K/MM, Hilton Diamond, Marriott Something, IHG Gold, Hertz PC, Avis PC
Posts: 8,156
You hear about the situations when the human fails, because they frequently end in drama and/or death. You never hear about the situations when the computer fails, because the "backup computer" sitting in one of the two seats takes over and almost always handles the issue without incident.
AF447 gets blamed on human error, but it was the computer that errored first and caused the human to take over. Of course, you could argue that if the computer had the capability to try and resolve the situation rather than just 'give up' so easily then it might have done a better job - and in some situations that may be true - but in other situations it's a stretch to think that would be true (eg, someone else has already mentioned US1549)
Mar 28, 2015, 9:46 am
#37
Join Date: Dec 2009
Location: RDU
Programs: DL DM+(segs)/MM, UA Ag, Hilton DM, Marriott Ti (life Pt), TSA Opt-out Platinum
Posts: 3,226
But then the FO panicked, ignored the stall warning (which can be heard repeatedly as they plummet) and caused the crash. When the pitot tube iced up, control could have been handed off to pilots on the ground, who would've used other indicators to control the airplane (just like the pilots on-board should have done).
AF447 was a result of multiple failures, but pilot error was most definitely the primary cause. It's sad because the Captain finally realizes that the FO's error is going to kill them right before the end of the CVR.
Really good read on it here: http://www.popularmechanics.com/flig...e-447-6611877/
Last edited by HDQDD; Mar 28, 2015 at 9:54 am
Mar 28, 2015, 7:15 pm
#39
Join Date: Dec 2009
Location: YUL
Programs: Aeroplan, Air Miles
Posts: 20
Given some of the issues that unexpectedly disrupt USAF and CIA drone operations, I'm not so sure about that.
Also, adding too many cooks to a kitchen may not generally equate with consistency of output and a lack of other operational problems/risks.
I'm certainly a technophile, but I don't adopt the mantra that any and all technology is inherently a great solution to the needs of the time. Automation is not perfect, and the results of automation aren't always just what we think they are or will be.
Also, adding too many cooks to a kitchen may not generally equate with consistency of output and a lack of other operational problems/risks.
I'm certainly a technophile, but I don't adopt the mantra that any and all technology is inherently a great solution to the needs of the time. Automation is not perfect, and the results of automation aren't always just what we think they are or will be.
Mar 28, 2015, 11:25 pm
#40
Join Date: Nov 2014
Posts: 164
Technology is an augmentation to humans, not a replacement.
You can code problemsolving into systems, but problem solving is limited to the problems you can list and provide a problemsolving algorithm for. When the problem is outside the boundaries of the known problems and solutions, the system breaks down.
Aircrews are constantly adjusting to unexpected issues (weather, winds, air traffic conflicts, equipment failures). Aircraft crashes are always caused by an accident chain, and in the last couple of decades technological failure almost always being a part of the failure. You need humans to deal with the failures that the system can't foresee, or deal with.
What happens with an electrical fire requires shutting down essentially all systems and handflying a visual approach?
Right now, problems are limited to a single aircraft. Automation and ground control risks entire fleets of aircraft. We have seen ATC failures that covered whole regions of the US. We have seen space storms that degraded satcoms and GPS. When a controller fell asleep at Reagan, the pilots just landed.
How would a fleet of automated and ground coordinated aircraft deal with a system failure dropping enroute ATC in a couple or three states?
You can code problemsolving into systems, but problem solving is limited to the problems you can list and provide a problemsolving algorithm for. When the problem is outside the boundaries of the known problems and solutions, the system breaks down.
Aircrews are constantly adjusting to unexpected issues (weather, winds, air traffic conflicts, equipment failures). Aircraft crashes are always caused by an accident chain, and in the last couple of decades technological failure almost always being a part of the failure. You need humans to deal with the failures that the system can't foresee, or deal with.
What happens with an electrical fire requires shutting down essentially all systems and handflying a visual approach?
Right now, problems are limited to a single aircraft. Automation and ground control risks entire fleets of aircraft. We have seen ATC failures that covered whole regions of the US. We have seen space storms that degraded satcoms and GPS. When a controller fell asleep at Reagan, the pilots just landed.
How would a fleet of automated and ground coordinated aircraft deal with a system failure dropping enroute ATC in a couple or three states?