Pilot here (learned to fly at age 16). There really should be a system for allowing air traffic control to remotely control flights en route, so pilots can sleep, read a book, bang their favorite stewardess/steward, etc.
The only phases of flight where you really need humans at the controls are takeoff, landing, and uh-oh what the fuck is going on (this being a broad "catch all" category which encompasses everything from an exploded engine and hydraulic failure, see United Airlines / Sioux City, to a medical emergency).
I would never fly on a pilotless airliner, but I'm fine with the pilots being asleep at the wheel during the en route phase of flight. There really isn't much to do except look at the GPS moving map and say "yep, we're over Iowa...... yep, we're still over Iowa". Course/altitude adjustments could all be beamed in from ATC.
I'd prefer them to be rested up for landing, or in case something unexpected pops up.
So presumably you would have an alarm that wakes up the pilot if this happens. Here's the problem: humans wake up pretty slowly. For 10 minutes or so after you wake up, you might as well be drunk as far as your reaction time and judgment are concerned. I don't know if this has been tested in high stress situations, and I would guess that the "warm up time" is faster if your adrenaline is pumping, but it's still going to be a lot worse than having already been awake.
There are two pilots. So assume they sleep in shifts.
The point is that the several hours en route are a good time for the pilots to rest. Takeoff and landing are the riskiest phases of flight. I'm mainly concerned that the pilots are on top of their game then.
I wake up with a hit of adrenaline a couple times a week (thanks to Plectron tones going off, indicating an emergency call of some nature).
I'd say that 'might as well be drunk' period lasts around 30 seconds for me (the process of getting dressed is generally just a foggy blur...). By the time I'm moving toward the truck, I'm more or less firing on all cylinders.
30 seconds is still an awfully long time in a plane when stuff is going sideways...
This is true, but the chance of an incident occurring during neither takeoff or landing, when both pilots happen to be asleep at the same time is fairly low.
On the other hand, sleep deprivation (which as the article mentions, can be far worse than being legally drunk) is common and pilots probably often perform the most difficult parts of the flight while under its effects.
I would think the risks of pilots sleeping (and, very rarely, getting caught unprepared for an emergency) would be minimal compared to the risks of pilots being regularly incapacitated by fatigue during the most difficult parts of the flight.
I disagree with the other pilot (but I'm only PPL) on his views about sleeping during boring stages of flight. Say for instance you've piloted badly into a CB, you suffer a freezing on your pitot static systems. The autopilot disengages.
If you were asleep, it might take a while for you to get your act together, co-ordinate to ascertain who has the flight controls and start CRM procedures.
If you were wide awake, thinking how to solve this tricksy problem, you might do a better job as your mind is already warmed up!
Personally, it takes me a long time to disengage my mind from "this tricksy problem" and bring it back around to real life. I suspect this delay is comparable to that for waking up with adrenalin. Anyway, no matter what, the policy is probably going to be that at least one pilot is awake with their mind on the job at all times.
It'd make sense for a sleep schedule on flights over 5 hours, let each pilot get an hour sleep after an hour or so in. That way both are rested well before they need to worry about landing.
('bang their favorite stewardess/steward', wait, that still happens? I suppose it could be fitted around a sleep schedule)
Hopefully this would also stop pilots interrupting the films every five minutes to tell you what the weather is like and how high the plane is etc etc.
YOU'RE A GLORIFIED BUS DRIVER. JUST GET US WHERE IT SAYS ON THE TICKET.
I'm only moderately experienced (1200 hours or so, single and multi-engine rated). Put me in any airliner with all systems working (dead crew, no mechanical faults, weather above minimums and sufficient fuel), and I'll get the thing to a safe landing 999 out of 1000.
I can't start it; I might not be able to taxi it in to the ramp; I surely can't diagnose system faults on an unfamiliar plane, but it's easy to get it to the ground and generally easy to get it stopped (especially if you choose an 8000+ runway, rather than the 5000 foot that might have been the original destination).
The vast differences don't matter much when everything is working and there are plenty of margins and you only have to land/stop.
I hate when people equate being at the helm of a majestic craft to being a bus driver.
There's far, FAR more akin to sailing than driving when you fly aircraft. Fluid dynamics are very similar, regardless of whether it's water or air you're sailing through.
I don't think being a sailor has ever been a shameful profession; much the opposite, in fact.
Noone was saying anything about shameful professions; but, just like if I was a passenger in a sail-ship, I wouldn't like the ship captain to interrupt whatever I'm doing in my cabin to state what is the wind direction today; I wouldn't also like the air-ship captain to interrupt the movie with the notification about airspeed and air temperature up in the sky. Let him worry about the majestic flying, and let the passengers mind their own business.
4) There is no economic incentive to have pilots fly less and sleep more
Since there will always be a tension in allowing pilots to sleep more and work less (thus costing airlines more money) we have to conclude that pilots are going to fly without sufficient sleep.
To make the system more resilient to failure, we should take the ad-hoc napping that pilots do now (which is undeniably adding some safety via fatigue reduction) and add a blame-free procedure around it to ensure safety (e.g. when one pilot goes to the bathroom the other one puts on their oxygen mask in case of explosive decompression; something similar should be done for the awake pilot.)
Yes, some models of the B777 (at least) will sound a warning if no controls have been moved/activated for 15 minutes:
"CREW ALERTNESS MONITOR
The FMC continuously monitors the activation of switches on the mode control panel, EFIS control panel, display select panel, CDUs, and VHF/HF push-to-talk (PTT) switches. When a predefined time elapses after the last control activation, the EICAS advisory message PILOT RESPONSE is generated.
The PILOT RESPONSE message can be cleared by pushing any control on any of the monitored systems or panels. If there is still no response after a short time, the EICAS caution message PILOT RESPONSE is displayed. If there is still no response, the warning message PILOT RESPONSE is displayed. Any control activation on the MCP, EFIS control panel, display select panel, CDUs, or VHF/HF transmitters resets the PILOT RESPONSE message. The PILOT RESPONSE message is inhibited at all altitudes below 20,000 feet while the aircraft is in a climb and the flaps are not up."
15 mins seems like a lot of time for a timeout. The B777 top speed is 950 kmph, which means the plane can travel ~237 km before someone wakes up. That's scary.
You seem to think that keeping the pilot awake is the only option (based on your car example and rhetorical question) but actually there are several options but it's important to be tracking pilot drowsiness as a datapoint first.
While Toyota's apparently monitors the driver herself, it seems most of the others on that list monitor driver input. Especially on a long haul flight where autopilot is on, they aren't likely to have any direct input, but should be monitoring the radio, engines, destination, and other key systems. The real concern with sleeping pilots is something like this happening [1], for which I never found a explanation other than sleeping pilots as particularly likely.
True, but until you can quantify exactly how much more likely a tired pilot will be to crash the plane, it's impossible to quantify the cost. It may be impossible to accurately measure.
> There is no economic incentive to have pilots fly less and sleep more
Crashes are tremendously bad for business. Also, the laws about pilot rest are incredibly strict. That doesn't mean that pilots won't break them, but it does mean that the "economic incentive" aspect is pretty much moot.
Trouble is, airlines have always been a strategic asset to governments, a situation the airlines play like a fiddle. So even if one is willing to use a crash as a "market signal" it is likely to get washed out by TBTF.
The same incentives apply to long-haul truckers and international passenger bus drivers - but there somehow regulation works, and they're not allowed (at least in EU) to drive nearly as much as airline pilots, due to the same fatigue concerns.
It's trivial to deny airlines the possibility of scheduling pilot shifts with '5 hours sleep in the last two days' as in the original article - the regulators simply have to do it, and as it will apply to all airlines, then even the low margins will accomodate the extra cents needed for less flexible scheduling of pilots.
I've often thought they should get rid of pilots from Aircraft, and fly them by autopilot. The number of accidents caused by human error is terrifying.
But also have a multiply redundant remote control system for emergencies the autopilot can't handle. So if something crops up, a crew of pilots trained to deal with nothing but emergencies can take over remotely.
It's interesting how flying a plane is far more challenging for a human than driving a car, but in the same way that solving a calculus problem is far more challenging than reading the subtext of a conversation: It's totally flipped for computers, because the former is about making adjustments based on a lot of interrelated numbers while the latter is about identifying signals and making lateral connections to respond.
It might be fair to say a human has to think a lot like a computer to fly a plane, while a computer has to think like a human to drive a car.
> The number of accidents caused by human error is terrifying.
Problem is that we don't know about the number of accidents avoided thanks to human interventions. Also, remote pilots will never be sure about the reliability of information transmitted from a failing aircraft.
I wonder if an autopilot would have landed the plane into the Hudson river, thereby saving everyone's lives? Something tells me no. If you have humans on the plane, you will need to have humans flying the plane. Yes! We should automate more to enhance the human's cognitive abilities, but no, we should not remove the human element. The human mind is still the biggest supercomputer we have when it comes to improvisation.
On water? It needed every bit of that pilot's skill, which also meant every bit of feedback about the situation was needed. No freakin' way - there's no bandwidth for that amount of information.
Forgot to mention, I was a UAV Pilot of the Shadow 200 TUAV in the US Army and logged about 480 combat flying hours. Here is a pic of me in a Shadow Ground Control Station in 2002ish. As a result, I know a good bit more than most about UAVs, which most people refer to as simply "drones".
A lot of accidents are avoided by human intervention, and often very creative "not in the book" intervention. For example, I know of one L1011 that got stuck in a nose up attitude, and the pilot regained control by having the passengers all move forward.
Furthermore, pilots are far, far better at detecting that a problem even exists. There's a heluva lot of information you can get from sound, vibration, smell, the 'feel' of the stick, view out the window, etc. Heck, even going back into the cabin and eyeballing the wings and engines have saved airliners from disaster. A Concorde prevented a crash this way.
Software, on the other hand, usually fails to recognize that anything has gone wrong, and pretty much never is able to correct itself, and certainly won't think of anything creative.
re: reliability of information... The same can be said for pilots sitting in the cockpit (see Air France 447).
In some ways, having the pilots outside the plane would be an advantage in an emergency. It's much easier to trust the gauges (which is almost always the right thing to do) when you aren't experiencing any of the physical sensations in the cockpit. There have been countless cases where pilots trust their own sense of up/down/left/right (in zero visibility conditions), and ignored the relevant cockpit indicators, to disastrous result...
In scenarios where gauges are (i) wrong or (ii) unable to communicate with a remote operator, aircraft are placed at significantly more risk of accident; sufficient to warrant human intervention for even trivial tasks. Putting the pilots on the ground turns the risk of an accident from pilot misjudgement in this situation into near certainty of an accident.
As someone working in this domain, I absolutely agree. Piloting an aircraft is also one of the riskiest jobs in the US [1]. Significant progress was recently made in fully autonomous flight [2].
Pilots are already expected to deal with all sorts of emergencies.
Will never happen:
0. Guidance signal loss
1. Guidance signal jamming
2. Guidance signal hijacking
3. Pilots on the ground don't have a vested interest with passengers, so the survival instinct won't necessarily kick in because they're not there.
4. Cost of mitigating all of the above would be much more than the risks of such a drastic change.
Could have been a good article promoting awareness about horrendous quality of life for most air crew but the editor just couldn't resist the horrible title and cliches about dangerous humans interfering with autopilot systems.
Automated flying isn't as cut and dry as is commonly thought, or as some of the other comments might indicate. Human error is definitely a problem and it's compounded by the realities of pilot working conditions which, unless you're a senior pilot at a major airline, means long hours and low pay. Pay (and most hours counted by regulations) is usually only counted from push-back to the gate and doesn't count hours pilots have to spend commuting, briefing, etc. This is difficult to change with the weight airlines have with the FAA and an industry where an unlimited supply of people who will suffer through anything to fly for a meager living, or in some cases even pay to work (see: "Shiny Jet Syndrome").
Autopilots fly and land extremely accurately [in the US at least, in visual conditions most landings are hand flown, and 'autolands' are done just enough as to keep up with periodic test regulation], but aren't feasible for a complete airline flight. I'm not convinced there is an automated flight management system with sufficient robustness to handle all the contingencies that have to be accounted for on a passenger flight. Drones are a completely different story. They're expendable, and get performance savings from not having life support. An airliner needs a pressurized cabin anyway, so you might as well shove a few pilots inside. And what if the guy remoting into the plane from North Dakota falls asleep? Or the network goes down, for that matter. How long before onboard autopilot figures that out, and has the plane going into survival mode looking for an airport to land (better hope they had great QA), and creates cascading delays throughout the airspace system. Or, hate to say it, another earthquake-tsunami occurs and takes out a couple airports, leaving dozens if not hundreds or airplanes looking for alternates within range. Now you're talking about a total integrated ATC automation. I'm just not convinced we're there yet, or it's coming any time soon. Leave the pilots be and give them some sensible rest/fatigue regulations instead.
The best analogy I have seen is that autopilot isn't flying the plane any more than robots in a modern operating room perform surgery while the surgeon sits back and takes a nap. Maybe they'll write an article about how that might be safer for us tomorrow.
Disclaimer: pilot. Also, I stole a lot of my information from Patrick Smith, who blogs frequently and crusades against common air travel misconceptions on http://www.askthepilot.com/
I made a comment a few days ago basically saying that the reasons that pilots don't do dumb things as opposed to surgeons is because their life is on the line if they mess up.
Plenty of pilots of smaller planes do dumb things. They usually get filtered out of the system if they try to become airline pilots, but there's plenty of room for dumbness in small single-engined aircraft too, and their life is just as much on the line, if not more so, than in an airliner.
A lot of dumb pilot stuff comes from overconfidence or a feeling of "it can't happen to me".
As the article notes, taking a nap while flying on autopilot is extremely safe, and definitely much safer than ending up completely blown during landing.
"No aircraft in the history of aviation has crashed because a pilot has gone to sleep at the controls," says David Learmount, a former RAF Hercules pilot and safety editor at trade magazine Flight Global.
I'd potentially trust a well designed autopilot a lot more than I trust the human pilot for routine tasks. As long as there's some sort of alarm for things changing significantly, I don't really see the problem if they want to sleep half the flight - just, you know, wake them up and give them a coffee before landing.
Autoland is already quite safe. However it requires special procedures on the ground to ensure the ILS signals are of sufficient quality, and so it is normally only used either in inclement weather where visibility is to low to trust human reaction times, or as a monthly system test with a pilot ready to take over in conditions suitable for a fully manual landing (in which accuracy of the ILS radio signals cannot be assured as ATC is not making sure other planes stay sufficiently far away from the path of the radio signals).
As daigoba66 says, air traffic control (the equivalent of which is a large part of driving) is still done by human beings.
Two other points.
1. Driving happens "in 1-and-a-bit dimensions" (i.e., cars are on roads, which are 1-dimensional plus a bit of width) and aviation happens "in 2-and-a-half dimensions" (i.e., planes are in the air, fairly close to earth, but with quite a lot of flexibility as to height), which means that there's far, far more space for planes than for cars.
If all the world's drivers were completely unable to tell where other vehicles were, all the roads would be filled with crashed cars within seconds. If all the world's airline pilots were completely unable to tell where other planes were, the effect would be minimal. Except that there'd be a bit of danger at takeoff and landing -- which, not altogether by coincidence, are two other things that are best done by human pilots rather than unaided machinery.
2. Cars operate in a world where there are also pedestrians, buildings to crash into, etc. Planes, not so much.
It seems to me that, to a very good first approximation, none of the problems self-driving cars need to solve apply to planes in sufficiently similar ways for "we can do it with planes, so why not with cars?" to be a good argument.
I've seen this referenced a few times in this article. Did the pilot consciously attempt to "land" in the Hudson River, or was the plane simply going down in the Hudson and the pilot was able to correct its pitch enough to land? If its the latter, I would say the autopilot could have done it better.
The pilot was directed to a nearby airport, but realized he could not make it. He then looked out the window and evaluated his options on where to land, and selected the Hudson.
There's nothing 'simple' about landing an airplane in water that is not designed for it, especially a glider.
Hehe...the Hudson airplane did not land in the river by default. The default choice in New York would have been to take out a few city blocks. This incident was a textbook example of good decisionmaking under pressure.
Most computer guys don't realize this, but you can't save the state of a flight. Most decisions are mostly permanent once made. And some, like for instance all decisions made on an aircraft with no engines, are completely irreversible. So if you don't make the right choice on the first try, you have to live with the wrong choice. At low altitudes under visual flight, a pilot makes multiple such decisions a minute. In an emergency, this number would be a lot higher; on the order of once every few seconds.
The only phases of flight where you really need humans at the controls are takeoff, landing, and uh-oh what the fuck is going on (this being a broad "catch all" category which encompasses everything from an exploded engine and hydraulic failure, see United Airlines / Sioux City, to a medical emergency).
I would never fly on a pilotless airliner, but I'm fine with the pilots being asleep at the wheel during the en route phase of flight. There really isn't much to do except look at the GPS moving map and say "yep, we're over Iowa...... yep, we're still over Iowa". Course/altitude adjustments could all be beamed in from ATC.
I'd prefer them to be rested up for landing, or in case something unexpected pops up.