Airbus turns to robots for in-flight emergencies

The Wall Street Journal reports that Airbus is planning on turning the controls of its planes over to Skynet friendly onboard computer copilots in case of emergencies like an impending midair collision. Currently, pilots are trained to turn off the autopilot when they encounter an emergency and maneuver the plane to safety themselves. But Airbus thinks that pilots sometimes overreact in such situations, unneccessarily shaking up the passengers (at best). The company plans to install the system on its A380 jet as early as next year and eventually install them on all their aircraft. Pilots, not suprisingly, are none too pleased with the move; Air Line Pilots Association safety offical Larry Newman says it's leading to pilots getting further and further away from the process of responding to emergencies themselves (well duh). Not to mention the whole, you know, robots making decisions that could directly affect hundreds or thousands of human lives thing. For its part, Boeing has said it will continue to rely on human pilots in case of emergencies.

[John Hughes @ May 27th 2006 11:43PM]

Today's planes cannot fly without computer help 100% of the time, can they?

[Johnny @ May 27th 2006 11:46PM]

This is an interesting development, but it is more of an incremental step than a sea change. Passenger jets have been equipped with a collision-avoidance system called TCAS for a while now. The classic example of a TCAS resolution is when two planes are on a collision course, the TCAS in one plane will instruct the pilot to "climb, climb," while the TCAS in the other plane says to "descend, descend." Pilots are required to obey TCAS commands -- they take precedence even over air-traffic control.

More information on Wikipedia:

http://en.wikipedia.org/wiki/Traffic_Collision_Avoidance_System

[Little Joe @ May 28th 2006 12:00AM]

Wow... what if it malfunctioned?

I mean imagine being a pilot and taking your hands of the wheel to let the computer do its thing... only to watch it lead you straight in to certain death.

eek.

[Tony Rayo @ May 28th 2006 12:18AM]

I know that planes can fly themselves (auto-pilot) while in air but I believe the take-off and landing is always done by a human pilot. There are very advanced data markets along runways though so I'm sure it'd be possible to have a plane take off, fly a course and land on it's own. Personally I have a lot of faith in the auto-pilot system and from what I know so do pilots/the FAA/etc but it's very hard for most people to hand over the keys to a machine 100%, and for good reason.

[Jeff @ May 28th 2006 12:18AM]

I am a Commercial Pilot and Flight Instructor and I'll throw in my two sense. Modern commercial planes are highly computerized, but the pilot has the final call. Taking that privilidge away is something Airbus has started to do to a point whereas it will not let the pilot exceed certain criteria not matter what(ie the video of the plane flying straight into the trees thats floating around). This has caused a lot of pilots I know to refuse to fly their equipment. TCAS advises pilots what to do but does not do it for them. What Airbus is proposing is completely whacko. The day a computer decides what to do is the day I don't fly on or near that plane anymore. Airbus systems are great but they don't always work. A friend of mine used to fly for British Airways and once had an Airbus plane refuse to start the engines, because the computer got confused and thought it was already running and in flight. While that did not cause a problem it brings up the question of what would happen if the plane were in the air, and it decided it was on the ground and turned the engines off. Airbus is a good company but their philosphy of taking the highly trained flight crews out of the equation is going to ruin them.

[Ryan Green @ May 28th 2006 12:24AM]

Hey, I remember when Airbus did a demo of this tech back in the day. A pilot was coming in for a landing at an air show and the onboard computer said that he was being a silly idiot and locked up the controls, doing what it thought best. The plane flew into a forest and had a nice fire break out.

I still trust an experienced pilot more than a computer these days.

[Special_K @ May 28th 2006 12:38AM]

Pilot error is still a huge reason for disasters in the aviation industry. As pertains TCAS, it only works when both pilots obey the instructions given to them by the machine. There was a middair collision (known as the Uberlingen disaster) when an air-traffic controller told 2 flight crews one thing, and TCAS told them another. One flight crew followed the ATC's instructions (which is against protocol) while the other followed TCAS' instructions (the way they're trained/regulated to do). The planes both ascended, and crashed. If TCAS had taken over, many lives (need I mention the fact that one plane was full of /INNOCENT CHILDREN/?) would have been saved.

[Adam @ May 28th 2006 1:47AM]

And thus the debate of Boeing vs. Airbus is finally sealed. The airbus philosophy of putting the pilots out of the loop will not, and should not be accepted by the flying public. No matter how many senarios are imagined by the engineers, they will never, ever, be able to predict every outcome of possible events. Furthermore, a computer is a component that can, and does, fail and I can't see how it can even be fathomed to have them in full control. To the above poster, most of the time the autopilot is engaged when the plane is about 400 feet above the ground or so, though a computer does control the throttles at take-off. It is then left on until around 50' or so when the pilots disengage the auto-pilot and hand fly it down to the runway. There are certain aircraft that can land with autopilot on certain runways and they do use it all the way to touchdown where the auto brakes deploy and stop the plane. The only real pilot involvement is to deploy the reverse thrusters and steer the airplane down the runway.

[Ben Hobbs @ May 28th 2006 2:37AM]

I think they've got it the wrong way round, the computers should fly the plane completely, with the pilot only interrupting in case of an emergency.

#5 Jeff

Its two cents, as in the money - I take it high-school is optional for Pilots, another reason to let computers handle it all.

[docarrol @ May 28th 2006 3:05AM]

#4
Actually, fully automated take offs and landings are not only possible, but common. I've read that the airlines require an auto-landing every X number of cycles. The auto-land feature is actually mandated in heavy fog and other zero visibility conditions. Keep in mind though, that this is only possible with a certain limited category of advanced autopilots (CatII/CatIII, and they have triple redundancy for safety), is only possible at airports with the latest and greatist systems to provide ground-side information to the planes (ILS), and the human pilots must be checked out on the systems so then know when and how to take over correctly.

A quick Google check for reference gives us:
http://answers.google.com/answers/threadview?id=482344
http://en.wikipedia.org/wiki/Autopilot
http://www.airliners.net/discussions/general_aviation/read.main/1882971/

[He Who Likes Pie @ May 28th 2006 3:08AM]

Heh heh, if things keep going this way we'll have the Terminator all over again. We'll know we're in trouble when nude tough guys start appearing from the future

[Johnny English @ May 28th 2006 4:14AM]

It's affect, not effect.

[diulei @ May 28th 2006 4:52AM]

How about someone invent some auto-driver technology for the zillions of crappy drivers out there, in their 93984398 cylinder SUVs, yapping away on their phones, oblivious to the rest of society? I swear sometimes a virus ridden PC would drive better than the people I see on the road everyday.

Seriously though, how do they test these kind of things? Just let go of the controls and hope your programming works?

[Murray @ May 28th 2006 5:31AM]

I Also am a pilot.

There has been more than one jet airline crash where the poor design of an Airbus system has been a major contributing factor.

Where the system itself may be sound, software is required for the system to operate. This requires developers, QA etc. The consequences of a poorly implemented design, a poorly written specifications document, poor programming or poor quality assurance processes can be grave.

IMHO, there are too many potential areas for a system like this to break down. A highly trained pilot will always have the edge.

#9. Autoland systems still require minimum acceptable visibility requirements for Instrument flight, a pilot babysits the system and is required to intervene if necessary. Additionally, very few airports have navigation aids certified to the required standard for autoland systems to be used all the way to the ground. The pilot always hand flies the final moments of the final approach and touchdown in the vast majority of cases.

[Todd @ May 28th 2006 6:47AM]

I congratulate our robot overlords' latest achievement.

[Francis Faludi @ May 28th 2006 7:36AM]

This is not original, just something I heard...

" An Airbus was designed if idiots to be flown by geniuses, whereas a Boeing was disigned by geniuses to be flown by idiots".

Note the final comment in the article "For its part, Boeing has said it will continue to rely on human pilots in case of emergencies." That sort of agrees with the first quote.

Another pilot line was (I too used to fly B 737's)... "If it ain't Boeing I'm not going".

I love automation, it made me look good in the days that I did fly but when all hell breaks lose I think I would like to have the option of dealing with it as I see fit and accorrding to my training and not hope that the latest software revision for whatever automated system Airbus wants to impose in this case, will fly me out of trouble.

[Francis Faludi @ May 28th 2006 7:40AM]

small correction to my post (a typo)

" An Airbus was designed BY idiots to be flown by geniuses, whereas a Boeing was disigned by geniuses to be flown by idiots".

[exst @ May 28th 2006 9:56AM]

Here's how a computer flies:
http://www.youtube.com/watch?v=Jg1uyRJjqaQ

[Aaron @ May 28th 2006 11:23AM]

I will never be afraid of airplanes. You will not find safer machines anywhere in the world.

[OSX @ May 28th 2006 11:26AM]

This is good news.

[Jono Cono @ May 28th 2006 12:25PM]

There are pros and cons for both, whether you rely on a computer or on human judgement, neither are perfect!

[Richard rountree @ May 28th 2006 12:45PM]

skynet funny

[tobjac @ May 28th 2006 12:46PM]

The video that Jeff, Ryan Green and exst is refering to is a urban legend. A computer did not fly that Airbus. Read the accident transcript in the link to find out what really hapended to the plane.

http://aviation-safety.net/database/record.php?id=19880626-0&lang=en

[Thierry @ May 28th 2006 1:08PM]

I'm an airline pilot flying the Airbus A-340. TCAS is a very effective safety feature in today's crowded skies. The computer commands need quick, smooth and accurate responses. Pilots have injured unstrapped passengers, by overreacting at the controls. Better training has improved the pilot's skills, but it makes a lot of sense to let the computers handle the airplane while in a potential collision situation. Pilots would still be monitoring the aircraft flight path, just as they do for a low visibility landing.

[David Reynolds @ May 28th 2006 1:20PM]

First of all, I'm an Avionics Technician in the Navy, and have worked on E-2C's, EA-6B's, and SH-60B's. I thought I would put in my two cents also.

There are very few modern aircraft that don't have an AFCS/EFCS (Automated Flight Control System). Basically, what that means is the pilot's control inputs never directly go to the flight surfaces, but instead go to a box which determines the "proper" amount of control surface movement. Yes, you can turn that system off, but then typically you also lose any input from the nav systems, and the stick/yoke becomes much heavier. Pilots do not like wrestling with heavy control inputs, and would rather have AFCS engaged. I know this because I tend to get my butt chewed when it goes down.

When it DOES go down, the pilot still has inputs, as the EFCS is an augmentation, not a replacement.

The earlier Airbus incidents were actually caused by the pilots configuring the systems wrongly. Pilot error IS the main cause of aircraft accidents, whether we want to believe that or not. The fix for the Airbus aircraft was for the system to run double-checks on switch configurations, and "check the pilot's work." We call that "Button-Fscking" and I think it happens to anyone working with tech.

I guess what I'm saying is that pilot's will resist change, but at heart they will still use it and come to rely on it. There's nothing wrong with computer control, and in most cases is a whole lot better than overloading your pilots.

Computers do not Button-Fsck. Only pilots and engineers do that. :)

[Jeff @ May 28th 2006 1:36PM]

#22

Actually what happened from everything I have been taught and studied about the accident is the plane got stuck in what I believe they called landing mode. This is where the plane thinks "YAY I GET TO LAND" and doesn't really want to do anything else. When the crew applied power to Go Around, the plane refused to climb out because it thought "Hey you told me we are going to land, so we are landing". Granted there may have been a lack of crew training in figuring out how to kick it out of that mode, or learning that if you go below a certain point it goes into that mode, but I would prefer there not be any thinking involved to take over control. If I take the controls, I want the plane to hand it over to me, not tell me no.

The computers in the flight prevented the crew from climbing out of the collision, even with the autopilot off and Airbus's computer will prevent the crew from doing certain things. I don't care what plane you are flying, or what your engine speed is, (btw according to the report, 83% N2 is damn near full power) if you yank back on the controls the plane will go up. If you watch the video link above, the plane does not go up, ie the crew was not in FULL control.

[NeoteriX @ May 28th 2006 1:52PM]

I agree Todd,

I for one welcome our new airplane pilot overlords.

[Ben @ May 28th 2006 3:24PM]

#25: I think you don't know your aerodynamics. Hell I only fly model aeroplanes, and I KNOW that if I have a plane with bugger-all power on, at a massive angle of attack, and barely above stall speed, then I'm not expecting a heck of a lot to happen in a hurry when I apply power. At best it will mush around for a distance, before slowly climbing and eventually returning to level flight after maybe 20+ plane lengths.

The power takes time to translate to speed, and you need additional speed to generate more lift. The alternative was to drop the nose to get more speed, but they had zero room to do so. Once the landing gear touched the first tree, it would have washed off heaps of speed, and it was all over.

[Jeff @ May 28th 2006 4:20PM]

#27

I do know my aerodynamics. Speed doesn't climb you, thrust does, if i could draw you the pretty little diagram I would, but just understand excess thrust it what climbs an aircraft. If i trim an airplane out to say 100 knots, and add power, it will almost immediately begin to climb. I fly REAL airplanes, I have been trained on the Saab 340, Dash 8, and am currently type rated on the CRJ. I have been doing this for many years, I think I know a little more then a freaking RC pilot. Yes, RC planes don't handle the same. Why? You have a tiny little engine. This thing has a huge Turbo Fan engine. If he added 83% N2 trust me, if he pulled back it would have gone up, otherwise the airspeed would start to go through the roof. 14 degrees pitch up is not that high. Normal takeoff pitch is usually 15 degrees.

[OnA Pest @ May 28th 2006 7:34PM]

Wow, an r/c pilot giving his insight as to the controls of a real plane... Thats crazy, 4 or 5 real pilots vs an r/c pilot..

I drove r/c race cars when i was little, can i tell a Formula 1 driver how to drive their car now?

HOWARD STERN IS GAY!
Opie and Anythony always win in the end!
XM 202 weekday mornings from 6am-noon eastern time.

[Hardcore @ May 28th 2006 10:14PM]

Hmm - Which would you bet your life on (literally)? - The chance that a programmer working 80 hrs a week to meet a deadline has no bugs in his code (i know it's qc'ed, etc...just saying) or a few live pilots who have been trained for hundreds of hours in all types of situations. I, for one, will take the pilots thank you.

[Jesse @ May 28th 2006 10:39PM]

I'm all for computer controlled planes. If there was no pilot (at least in the plane) then we wouldn't have to spend hundreds of billions of dollars trying to prevent hijackings that almost never happen.

[Thierry @ May 29th 2006 11:58AM]

Actually, don't dismiss R/C pilots comments so easily; aerodynamics laws are the same for everybody! Poster 27 got it right: The pilot of that A-320 got his aeroplane in such a state of low energy, that when he decided to apply TOGA thrust (full thrust), the high drag at such a low speed (Alfa Floor) and the long time needed to achieved spool up meant that the computers could only prevent a disastrous stall and fly the aeroplane gently into the trees!
Captain Airbus A-340

[Lee Gleason @ May 30th 2006 12:53PM]

In the future, the cockpit crew will be reduced to one pilot and a dog. The dog is there to bite the pilot if he tries to mess with the automated flight controls. The pilot is there...to feed the dog.

[Osiris @ Jul 25th 2006 7:38PM]

No one has addressed the issue of hacking the command of the plane and thus controlling it wirelessly themselves, negating the need for a suicide terrorist attack..

Remember, nothing is hacker proof.

[ries @ Aug 1st 2006 4:03PM]

I think it's good that computer take over in part of the world, it happens everywhere around us. Telephone systems let you cann around the world... Nobody complains there is no operator anymore... Computers build cars and robots build... robots!!!
The problem with flight computers it just a huge factor more complicated. It takes time untill we humans will master it 100% and so the flight can be near 100% save.

David might bable to comfirm, plance have a limited number of inputs to learn about there surroundings. That's why it's so hard for them control at always 100%.

We will get there... and we will want it.. it just takes time to make it work 100%. BTW the airbus accedent happend in 1988. Now we are 18 years later.
We humans get smarter, computer calculate faster... Things WILL get better.

Ries

[Zach @ Nov 9th 2006 11:55AM]

From Meryl Getline's site (she used to write a column in USA Today)

Question 3: Can airplanes take off and land on autopilot?
The autopilot can fly an airplane once in the air, and land it, but it cannot be used during takeoff. Most pilots will hand fly the plane up to 10,000 feet or at cruise altitude before they engage the autopilot. The autopilot makes constant corrections for altitude and course. This frees us up to look out for traffic and other duties. The autopilot is mandatory for landings during foggy conditions or in a snowstorm when visibility is low. Restrictions prohibit landing using autopilot during very windy conditions.

It could be out of date, but I'm not a pilot.

[joe shmoe @ Nov 9th 2006 12:31PM]

The autopilot cannot guide the airplane down the runway centerline on a takeoff roll, but can be engaged quite quickly after takeoff.. There are limitations set in place by the manufacturer that dictate what they are. Regarding automatic landings, they can be performed regardless of visibility, but only under highly specialized conditions at particular locations by particular airplanes and crews.