Planes Without Pilots

HughPickens.com writes: John Markoff writes in the NY Times that in the aftermath of the co-pilot crashing a Germanwings plane into a mountain, aviation experts are beginning to wonder if human pilots are really necessary aboard commercial planes. Advances in sensor technology, computing and artificial intelligence are making human pilots less necessary than ever in the cockpit and government agencies are already experimenting with replacing the co-pilot, perhaps even both pilots on cargo planes, with robots or remote operators. NASA is exploring a related possibility: moving the co-pilot out of the cockpit on commercial flights, and instead using a single remote operator to serve as co-pilot for multiple aircraft. In this scenario, a ground controller might operate as a dispatcher, managing a dozen or more flights simultaneously. It would be possible for the ground controller to "beam" into individual planes when needed and to land a plane remotely in the event that the pilot became incapacitated — or worse. "Could we have a single-pilot aircraft with the ability to remotely control the aircraft from the ground that is safer than today's systems?" asks Cummings. "The answer is yes."

Automating that job may save money. But will passengers ever set foot on plane piloted by robots, or humans thousands of miles from the cockpit? In written testimony submitted to the Senate last month, the Air Line Pilots Association warned, "It is vitally important that the pressure to capitalize on the technology not lead to an incomplete safety analysis of the aircraft and operations." The association defended the unique skills of a human pilot: "A pilot on board an aircraft can see, feel, smell or hear many indications of an impending problem (PDF) and begin to formulate a course of action before even sophisticated sensors and indicators provide positive indications of trouble." Not all of the scientists and engineers believe that increasingly sophisticated planes will always be safer planes. "Technology can have costs of its own," says Amy Pritchett. "If you put more technology in the cockpit, you have more technology that can fail.""

Comments

Sensors wrong

[jfdavis668]

Many airline disasters are caused when sensors go wrong or the output is confusing. Air France 447 was one of those incidents. It would be even more confusing to a remote pilot.

Re:Sensors wrong

[Rei]

80% of accidents are pilot error. I think the further we move toward automation, the better. Getting the pilot physically out of the plane is a major hurdle that one has to step over in that regard.

Re:Sensors wrong

[aaaaaaargh!]

To add to this, people seem to forget everything that happened more than a month ago or so. I'd like to see the computer that would have ditched US flight Airways 1549 perfectly into the Hudson River just minutes after the start.

Re:Sensors wrong

[Anonymous Coward]

That figure must probably be read like "80% of accidents could have been prevented if pilots had reacted in exactly the right way when some unforeseen technical failure occurred ..."

Re:Sensors wrong

[TigerPlish]

AF 447 would've had a better chance if the idiotic Airbus un-overridable Flight Envelope Protection had not silenced the stall horn when the aircraft exceeded what the FEP thought was a valid angle of attack.

The high angle of attack *was* valid, it was reality, it was happening, and whenever the pilots would push the nose down to correct the stall, the stall horn would come on again, so they would pull the nose up again. Which *erroneously* silenced the alarm.

From the Wikipedia article:

The stall warnings stopped, as all airspeed indications were now considered invalid by the aircraft's computer due to the high angle of attack.[27] In other words, the aircraft was oriented nose-up but descending steeply. Roughly 20 seconds later, at 02:12 UTC, the pilot decreased the aircraft's pitch slightly, airspeed indications became valid and the stall warning sounded again and sounded intermittently for the remaining duration of the flight, but stopped when the pilot increased the aircraft's nose-up pitch. From there until the end of the flight, the angle of attack never dropped below 35 degrees.

You see the problem there? The plane thought for the pilot, and it thought wrong.

FWIW, Boeing's FEP can be completely over-ridden, but not Airbus'. Even with all the benefits of FEP, I think the pilot should always have final say.

Re:Sensors wrong

[Registered Coward v2]

To add to this, people seem to forget everything that happened more than a month ago or so. I'd like to see the computer that would have ditched US flight Airways 1549 perfectly into the Hudson River just minutes after the start.

A point very important you make. Automation is great for instances where sensor data is accurate and a proscribed course of action can be safely followed. More automation can be useful in such cases. However, it's the edge cases where the pilot's judgement is needed to safely operate an aircraft. The USAir 1549 is an excellent example, as is United Flight 232. Could a remote pilot glide a 767 to a safe landing, and avoid cars on the abandoned runway that the copilot happened to know existed, as happened with Gimli Glider? As with many highly complicated devices, automation is a great tool to help the operator in routine, and some casualty, operations; however when things go not as planned and a new twist is added to the scenario you need judgement, not rote rules, and judgement is sorely lacking in automation and difficult to do if you are thousand of miles away and only know what the sensor data feed tells you.

Re:Sensors wrong

[Mr D from 63]

>80% of accidents are pilot error

That doesn't tell the whole story. How many accidents were averted due to human intervention?

Re:Sensors wrong

[SuricouRaven]

It's not a technical issue, it's a policy issue: A computer will conclude 'no runways within range, you are all going to die.' A human can conclude 'Screw the procedures, I'm going to try a high-risk controlled crash into a river and hope for the best. No other options.'

Re:Sensors wrong

[Richard_at_work]

The decision tree to get to the idea of putting the aircraft down onto anything other than a runway with a CAT I, II or III landing system is beyond automation. Without ILS, the computers on board the aircraft had no idea what the ground infront of them looks like, other than "its there".

Re:Sensors wrong

[Richard_at_work]

The stall warning was cut off because the readings being fed to it made no sense (they dropped below absolute minimums - the reasoning being that the pilots having sat through 5 minutes of warnings and not changing their approach to flying the aircraft, it won't suddenly fix itself as the horizontal speed drops to zero) - it wasnt cut off because of any automation systems, it was cut off because the readings didnt make any sense.

But it takes more than an avid Boeing fan to actually read the AF447 report.

Re:Sensors wrong

[Rei]

You think he was steering that thing like the Wright Flier? The A320 is a highly automated plane. The decision to land in the Hudson was his, but everything else was fly-by-wire - yes, in glide mode too. You can't overrule it on the A320. Everything involved in keeping the wings level and nose at the gentlest possible angle that wouldn't stall was managed by the flight computer - which is why the plane didn't flip when it hit the water.

The chief engineer of the A320 project actually referred to the plane as "pilot-proof". If Sully had done absolutely nothing from the time the plane lost power, you know what would have happened? Apart from the plane heading to the Hudson, everything else would have been the same. The plane would have done its best to maintain its original trajectory and kept itself stable as long as it could without losing altitude. Then it would automatically have lost just enough altitude to keep it from stalling, keeping itself flying level for as long as physically possible. If Sully in such a situation had started jerking up on the controls trying to get altitude that the plane wasn't capable of achieving, you know what would have happened? Nothing. The plane would have ignored him.

The pilot in a modern plane like the A320 doesn't manage, and isn't even allowed to manage, the finer details of flying. The plane does that. The pilot is only there for general overarching strategy. The plane makes it happen.

Even when crashing.

Re:Sensors wrong

[oodaloop]

Why can't a computer figure out if there are no runways to fall back on emergency options like roads or rivers? Hard, I'm sure, but far from the hardest part of flying a plane.

Re:Sensors wrong

[michelcolman]

80% of accidents may be due to pilot error, but probably close to 99% of all would-be crashes due system failures do not turn into an accident because of pilot intervention, and therefore never make it into the accident statistics. Take the pilots out, and you'll see at least an order of magnitude more crashes unless technology improves drastically.

I'm a pilot, I've never had a crash (like the vast majority of pilots), but I've had several situations where automation failed (either completely shutting off or doing something unexpected and dangerous) and a crash would have resulted if we hadn't taken over.

In fact, there are lots of crashes that are attributed to pilot error not because the pilots were the only cause for the accident, but because some system failure occurred that should have been handled safely by a well-trained pilot and somehow wasn't. We are expected to handle these problems, so if we don't, it's our fault (and rightly so).

Take Air France 447 for example, airspeed sensors iced up, autopilot disconnected, other flight crews in the past had had the same problem but handled it well, these pilots got confused and crashed. Probably goes into the statistics as pilot error, but without pilots the plane would have crashed anyway. Every time, including on those flights where the crew did handle the situation correctly (even with inadequate procedures for this particular failure at the time) and landed safely.

Another example, the Turkish Airlines flight that crashed short of the runway in Amsterdam. The plane was flying on autopilot, yet it's "pilot error" because the pilots should have immediately reacted when the autothrottles pulled the throttles back to idle and the airspeed decayed rapidly. Caused by a malfunctioning radio altimeter which let the automation think the plane was low above the runway and it was therefore safe to pull the throttles back for touchdown. There's a reason why we have an initial training and a yearly recurrent training for automatic landings. Haven't had the training? Manual landings only.

So no, automation is not safer than human pilots. Not by a long shot, at least not yet. And given the slow pace of technological advancement in aviation, it will be a very long time before it will be.

Take military drones, for example. Their mission is not exactly complicated: in relatively nice weather, take off, fly a predetermined route, drop some bombs, fly back and land. There aren't nearly as many drones as airliners flying around, yet drones crashes happen all the time, it's not even news.

Re:Sensors wrong

[Richard_at_work]

You are talking about billions of sites across the globe - and sites would be dependent on weather, local conditions, change of use etc etc etc. What if on that fateful day, one of the Hudson's ferries happening to be in the way but there wasn't enough power available to allow the use of the forward radar? What if the field pre-chosen had suddenly turned into a camp site? Or had a combine harvester and a fuel bowser parked in the middle of it. What if the Hudson was iced over?

As for the APU, the issue is that it simply was way down on the checklist - and a lot of things on that checklist can't be done in parallel etc Just because a computer has a speed advantage, doesn't mean it can use it. It was a concious decision from the PIC to fire up the APU out of checklist order. You don't find concious decisions coming from computers.

Re:Sensors wrong

[brambus]

This is such an ignorant post I can't believe it. It appears you've never actually had an airplane's controls in your hands.
1) Fly-by-wire isn't what you think it is. It simply means there are no mechanical linkages.
2) Airbus' computer-over-human approach is no panacea and it has resulted in numerous near-disasters, one of the most recent ones.
3) Even Airbus isn't religious about this approach. Read up on Alternate Law and Direct Law.
4) Had Sully not maneuvered USAirways 1549, it'd have landed in the middle of housing.
5) Water landings require you to do a flare & float to stall just feet above the water level to minimize airspeed. If he had not done this, the airplane could have easily smashed itself apart, since an A320 power-off glide rate of descent is around 1500 fpm. Water isn't soft at these kinds of speeds you know.

We'll never learn

[Aethedor]

And when the first plane crashes due to a bug in the pilot software, we all start wondering again if removing the pilot was a wise decision.

This whole Germanwings plane crash shows, again, one important thing: people suck at dealing with risks. Several hundred thousands of flights went well. The last incident with a pilot causing a plane to crash was back in 1995. The Germanwings plane crash was an incident. We must learn to treat it that way, as an incident. No reason to panic and start changing policies, rules and procedures. With every change, new risks and new ways of things to go wrong will be introduced. When that happens and you again make changes, you end up in a loop of changing things. The result: the changes will cost a lot of time, energy and money while the risks are not reduced.

We need to start accepting that risks are part of our life. Unacceptable risks need to be dealt with, but more important: acceptable risks should be accepted, even when they occur!!!!

Race to the bottom much?

[johnnys]

Just put two reasonably competent people in the cockpit at all times and stop trying to f**k an extra penny out of every dime, you cheap chiselling b*st*rds.

Right now many feeder airlines are barely paying a living wage for their junior cockpit staff, so stop pretending that the personnel costs are going to put you out of business. You're certainly not passing along the recent fuel cost savings to us sardines.

(I haven't had my coffee yet, so that's my excuse for the "negative tone" in this post.)

Perfect security

[scotts13]

Anyone want to guarantee 100% perfect security for ANY wireless communication? Because if we have remotely piloted airliners (either because there's no pilot, or the pilot is suicidal) someone WILL hack into it.

Re:Perfect security

[meta-monkey]

Good point. Someone should recommend they not use home wifi routers for the mission-critical communications on commercial airliners.

Re:Technology can indeed fail

[BostonPilot]

As a pilot, I would never fly an aircraft which has a remote capability to take control away from me. As a passenger, I would never fly in an aircraft in which remote control could be taken away from the crew. I don't even think the "remote copilot" is a good idea. There are a lot of good reasons to fly with a crew of 2... it's not just workload and risk of incapacitation. When things get weird, it's good to have another person to bounce ideas off of. United Airlines Flight 232 is a great example of that, and there are plenty more.

This isn't the first time we've seen suicide by the crew, but it's extremely rare (I can think of maybe 5-6 in 40 years). It sucks to be the people in the back when that happens, but it also sucks when an airplane drops on your house and kills you and your family. I'm as worried about one than the other which is: not very. We can improve the situation some: better screening of crew, USA style "always two people in the cockpit" rules... To suddenly decide that we can't trust the crew despite the fantastic safety record aviation has is just ludicrous.

 

Re:And airplanes without passengers...

[sonamchauhan]

Instead of having airlines staff the flight deck back up again to previous levels (i.e. reinstating the flight engineer, or adding a third pilot), cue the snake-oil salesmen with the same pitch - irresistible to airlines - which got us into this mess.

"Shame about that last crash...
"You know, you _CAN_ increase safety _AND_ lower costs
"How? Well, for starters, one less pilot on the flight deck...

Horse, cart

[fulldecent]

90%+ of comments here have been regarding lack of onboard pilots with commercial passenger flights.

Naturally, the first offboard pilot flights would be with cargo only. And that is way more relevant and less sexy discussion.