Ryanair CEO Michael O'Leary: Who Needs a Copilot?

[Cyrus]

So if you didn't have an ILS system on board and you needed to land, what would you do?

Fly somewhere where you can see the runway and land there. You do realize that a CAT-II or III you can't even see the ground to flare - and you think a pilot can do this with a 'good gps and instrument?' Ooooo-kayyyyy.

 

[JaredJames]

If the airfield vectors you onto the runway and your instruments are set correctly, why could you not fly a raw data approach? Definitely not easy, if anything extremely risky, but if you (for whatever reasons) don't have the ability to leave the bad weather region (fuel limitations etc) then you have to do something.

 

[Cyrus]

If the airfield vectors you onto the runway and your instruments are set correctly, why could you not fly a raw data approach? Definitely not easy, if anything extremely risky, but if you (for whatever reasons) don't have the ability to leave the bad weather region (fuel limitations etc) then you have to do something.

If you have a working VOR, you can do a step down approach. But getting verbal vectors to a runway in IMC? ...shudderr...That's a death sentence . Thats why you have 45 minutes reserve fuel - to go somewhere else. It would take an absolute moron of a pilot to go somewhere with no alternate airport in bad weather!

You really think the tower saying "turn left, turn right, go up, now down" is going to help you here?

EgeT-F9-1KI&feature=related[/youtub...d pieces of your airplane over several miles.

 

[Andre]

If you have a working VOR, you can do a step down approach. But getting verbal vectors to a runway in IMC? ...shudderr...That's a death sentence .

In the old days when men power was still cheap, we used to do Ground Controlled Approaches (GCA) assuming that the standard ILS (instrument landing system) was too susceptible for jamming/interference. The GCA radar approach was cat II rated with a decision height of 200 feet and some 900 meters visibility. The guidance was done by a guy who would talk constantly - non stop, giving instructions for the last four five miles of the final approach, something like:

Drifting slightly left of course, turn right two degrees - heading 207, on glidepath, back on course; turn left one degree heading 208; getting slightly above the glidepath; ease her down, one mile to go; back on glide path; resume normal rate of descent; on centre line, heading good, approaching decision height; look ahead for the runway; etc etc

No problem, daily routine 25 years ago, regardless of zero or 400 passengers.

Anyway the youtube approach shows that we have the automated techniques to perform better landings than a pilot can do, just as easy as a drone can fly from A to B. But that doesn't mean that you can eliminate a pilot altogether like in the engineers dream, for the simple reason that you would eliminate airmanship and situational awareness which are the main tools of a superior pilot. Remember the definition of that

A superior pilot is a pilot who uses his superior judgement to avoid getting into situations that require his superior skills

.
Autopilots are not in that category, but you may wonder how many superior pilots are required to witness the autoland system doing its work.

 

[Cyrus]

Why are you under the impression that a pilot is more situationally aware than an autopilot? That's the whole reason why the FAA is moving over to the ADS-B transponder system in NextGeN. Pilots suck at situational awareness and need a GPS based TCAS to help them out. A computer can keep track of other aircraft thousands of times a second while flying the airplane, in fog, and optimizing a 4-D based trajectory - all at once.

Also, was your verbal IMC landings done to minimums like in this video? You said the ceilings were 900 meters, that's almost 3k feet. Huge difference.

 

[Andre]

Why are you under the impression that a pilot is more situationally aware than an autopilot? That's the whole reason why the FAA is moving over to the ADS-B transponder system in NextGeN. Pilots suck at situational awareness and need a GPS based TCAS to help them out. A computer can keep track of other aircraft thousands of times a second while flying the airplane, in fog, and optimizing a 4-D based trajectory - all at once.

Also, was your verbal IMC landings done to minimums like in this video? You said the ceilings were 900 meters, that's almost 3k feet. Huge difference.

Autopilots have no idea about microbursts and windshear, trucks on the runway, crosswind, other aircraft with emergencies getting priority, hail and lightning damage to circumnavigate. Making early decisions about diversions when headwind was stronger than forecasted and the fuel running lower than expected etc etc, but above all, hard pieces in the air, not seen on radar.

Maybe reread what I said about the limits: Cat II, Ceiling 200 feet, visibility 900 (later 800) meters.

 

[Cyrus]

Autopilots have no idea about microbursts and windshear, trucks on the runway, crosswind, other aircraft with emergencies getting priority, hail and lightning damage to circumnavigate. Making early decisions about diversions when headwind was stronger than forecasted and the fuel running lower than expected etc etc, but above all, hard pieces in the air, not seen on radar.

Pilots can't see mircrobursts or windshear either. What do you mean by can't see crosswinds? Autopilots can land in crosswind. All of the things you listed are easily done by a computer...in fact a computer already does most, if not all of it. The pilot enters in a destination and way points and the Flight Management System (FMS) computes a path.

Maybe reread what I said about the limits: Cat II, Ceiling 200 feet, visibility 900 (later 800) meters.

Ah, Ceiling 200 feet, my mistake.

 

[Andre]

Pilots can't see mircrobursts or windshear either.

But a superior pilot knows where they can be found, looking at the clouds.

edit: moreover if the airspeed suddenly increases, he will initiate a missed approach, going full power, knowing that he just entered the microburst, whereas the autopilot will decrease trust ad adjust attitude to reduce the airspeed again, and in doing so just made a fatal error leading to the crash.

What do you mean by can't see crosswinds? Autopilots can land in crosswind.

but the superior pilot has just decided that the cross wind gusts are above limits even if the steady state winds are well below.

But there is nothing to do for a autopilot if the balloon or glider is not on the radar.

 

[Andre]

Anyway, maybe compare it with driving in traffic, I don't think a lot of people fancy automatic driverless cars, perfectly capable of driving exactly where they need to go, but lacking anticipation of children running behind the shiny red balls that cross the road.

By the time that the automatic systems are capable of that kind of drivemanship / situational awareness, one may wonder what humans still are doing on this world.

And for those who think that they can make flying fully automatic, see if you can work in a flying company for a few months and evaluate all the incidents, near incidents, irregulaties, unexpected changes in situations, to see if an automatic system could have coped with that.

 

[Cyrus]

True enough Andre. Although, the technology is getting there quicker than you think .

 

[JaredJames]

I have to agree that so far as take off, fly the route, land goes it is certainly capable with current technology. It is the additional factors as andre has pointed out that make a human pilot 'better' than current computer systems. Does anyone know if they've conducted tests of pilotless drones in crowded sky conditions such as those found around a major airport? (Pilotless as in flying under its own control, not like the predator drones)

I think the human factor is more important when it comes to military operations as opposed to civilian, but none-the-less required.

 

[DaveC426913]

... see if an automatic system could have coped with that.

Strangely, there is a strong double standard when it comes to automated versus human systems.

If a human were flying when an accident occurred, the public will automatically assume she did the best she could but it wasn't enough. It is forgiveable.

If a computer were flying when an accident occurred, we will automatically assume the computer made a mistake and/or couldn't handle the sitation. A pilot should have been there to oversee it. This is unforgiveable.

 

[Cyrus]

I have to agree that so far as take off, fly the route, land goes it is certainly capable with current technology. It is the additional factors as andre has pointed out that make a human pilot 'better' than current computer systems. Does anyone know if they've conducted tests of pilotless drones in crowded sky conditions such as those found around a major airport? (Pilotless as in flying under its own control, not like the predator drones)

I think the human factor is more important when it comes to military operations as opposed to civilian, but none-the-less required.

UAVs are not allowed to fly in the NAS (National Air System) unless under very special conditions near unpopulated areas.

 

[JaredJames]

UAVs are not allowed to fly in the NAS (National Air System) unless under very special conditions near unpopulated areas.

What about a demo setup (abandoned airfield with simulated traffic, perhaps other drones)?

I suppose if there was a central system which monitored all aircraft and issued instructions, it could handle airspace separation far better than a human system.

I wouldn't expect them to fly a drone straight into LAX, but it would be nice to see some sort of test setup where they could actually try out an automated airfield system. In my opinion, having a central system issue commands to the aircraft autopilot, would give far better control of the airspace.

 

[Cyrus]

What about a demo setup (abandoned airfield with simulated traffic, perhaps other drones)?

I suppose if there was a central system which monitored all aircraft and issued instructions, it could handle airspace separation far better than a human system.

I wouldn't expect them to fly a drone straight into LAX, but it would be nice to see some sort of test setup where they could actually try out an automated airfield system. In my opinion, having a central system issue commands to the aircraft autopilot, would give far better control of the airspace.

Google NextGen. The current system is a combination of human controlled/ airspace automation. You can read more about ATC by looking for the FAA document JO 7110.65T.

 

[mugaliens]

I have to agree that so far as take off, fly the route, land goes it is certainly capable with current technology. It is the additional factors as andre has pointed out that make a human pilot 'better' than current computer systems. Does anyone know if they've conducted tests of pilotless drones in crowded sky conditions such as those found around a major airport? (Pilotless as in flying under its own control, not like the predator drones)

While UAVs such as the Global Hawk are perfectly capable of taking off, flying the route, and landing totally without human intervention, "see and avoid" still applies, which is why they're not allowed to fly in the national aerospace system, except under certain conditions. Even so, the Global Hawk has an operator connected to it via satlink and monitoring it at all times. The radios aboard the drone allow the operator to talk to ATC the same as any piloted aircraft, and the operator is required to be a certified IFR pilot.

Should the satlink ever fail, the Global Hawk is programmed to squawk 7600 or 7700, depending on the situation, broadcast its intentions to ATC, fly to the nearest suitable military airfield, land, stop, and wait for a tow.

I think the human factor is more important when it comes to military operations as opposed to civilian, but none-the-less required.

I agree.

 

[Andre]

I think the human factor is more important when it comes to military operations as opposed to civilian, but none-the-less required.

One more then. I think not.

Whilst the human factor, as in anticipating the child running behind the shiny red ball, may be much more present in military air combat scenarios, it's equally important to break off the landing procedure after picking up that fire truck on the runway, that inadvertenly was sent there to pick up the remains of that dead deer hit by another aircraft, due to miss co-ordination between the local tower air traffic controller and ground control.

 

[JaredJames]

One more then. I think not.

Whilst the human factor, as in anticipating the child running behind the shiny red ball, may be much more present in military air combat scenarios, it's equally important to break off the landing procedure after picking up that fire truck on the runway, that inadvertenly was sent there to pick up the remains of that dead deer hit by another aircraft, due to miss co-ordination between the local tower air traffic controller and ground control.

I'm not entirely sure why you quoted me on this one, I was referring to the fact that military missions, where the use of deadly force (missiles etc) may be present, a human being there assessing the action about to be taken is important (unaccounted variables for instance).
As you said yourself, the civilian side of things is simply A to B and as such those judgements are not required. So, in as far as human judgement is concerned, I'd say the military requires a human face in the cockpit far more than civilian. All of the problems you have pointed out for civilian also exist for the military, but not all the military situations are present for civilian airliners (or do 'fire engines scraping up dead deer' scenarios only occur at civilian airfields?).

You keep mentioning crew coordination as a reason to reduce crew numbers. In which case a purely computerised system is perfect as all instructions will follow the same scrutiny and procedures each time regardless of how much work the computer is doing and no distractions possible. It is simply a case of emulating human 'awareness' that is required.

A good sensor based system (cameras, motion tracking etc) focussed on the runway and taxiways would be able to alert any aircraft to possible hazards, in much the same way as TCAS works now (there's something on the runway, in this case a fire engine, go around). So your "shiny red ball" scenario could be avoided in most eventualities unless it was extremely sudden, in which case neither human nor computer could respond in time. (I understand it is significantly more complex than this, but the principle is sound).

 

[Andre]

I'm not entirely sure why you quoted me on this one, I was referring to the fact that military missions, where the use of deadly force (missiles etc) may be present, a human being there assessing the action about to be taken is important (unaccounted variables for instance).
As you said yourself, the civilian side of things is simply A to B and as such those judgements are not required. So, in as far as human judgement is concerned, I'd say the military requires a human face in the cockpit far more than civilian. All of the problems you have pointed out for civilian also exist for the military, but not all the military situations are present for civilian airliners (or do 'fire engines scraping up dead deer' scenarios only occur at civilian airfields?).

It actually happened to me a few decades ago, but if you'd want to express the importance of catching such an eventuality in number of potential casualties, should it have gone unnoticed, then obviously in a civilian scenario, it would have been much more important.

You keep mentioning crew coordination as a reason to reduce crew numbers. In which case a purely computerised system is perfect as all instructions will follow the same scrutiny and procedures each time regardless of how much work the computer is doing and no distractions possible. It is simply a case of emulating human 'awareness' that is required.

...about which I have said, that would wonder if this level of automation can ever be reached, as it requires anticipation of sometimes new possible scenarios. That's the same discussion if AI can ever reach the level of natural intelligence.

 

[skeptic2]

To take this another step, The Federal Railroad Administration has mandated a Positive Train Control system be in operation by 2015. This system will control speeds and stop the train if the engineer fails to do so and will be fully redundant. Already people are asking if, with that system in place, there really needs to be an engineer on board the train. It's not as if an engineer can avoid an accident by swerving or braking.

 

[JaredJames]

It actually happened to me a few decades ago, but if you'd want to express the importance of catching such an eventuality in number of potential casualties, should it have gone unnoticed, then obviously in a civilian scenario, it would have been much more important.

Couldn't agree more.

...about which I have said, that would wonder if this level of automation can ever be reached, as it requires anticipation of sometimes new possible scenarios. That's the same discussion if AI can ever reach the level of natural intelligence.

I agree regarding awareness, however, if you were to look at a system which monitored the airfield roughly as I described above, I don't see how anything 'new' could occur. If the system had a high enough resolution so far as watching the airfield goes (can pick out objects relatively small in size which endanger aircraft), it could keep track of all things moving around the airfield in much the same way radar looks to the sky (although significantly more complicated). In such a case, you could argue the monitoring system is 'aware' of everything on the airfield and capable of making decisions (although they would be mainly predetermined - stop aircraft taxiing, go around etc).

 

[Borek]

Speaking of coordination problems. How many computers are there in a plane and aren't they coordinated and switched off if they don't agree?

Or was it a space shuttle?

 

[JaredJames]

Speaking of coordination problems. How many computers are there in a plane and aren't they coordinated and switched off if they don't agree?

Or was it a space shuttle?

Yes aircraft have that system, I believe it's a 'democratic' style system where they compare the readings and go with the majority. e.g. 3 computers compare readings, if one disagrees it's ignored. A simplistic description but basically there.

 

[DaveC426913]

Yes aircraft have that system, I believe it's a 'democratic' style system where they compare the readings and go with the majority. e.g. 3 computers compare readings, if one disagrees it's ignored. A simplistic description but basically there.

Is this primarily for redundancy in the system? Or redundancy in the decision?

If they're running identical programs, under what circumstance would they disagree?

1] One of the computers goes wonky. Redundancy in the system ensures two others can take up the task.

2] Each computer is given differing parameters of decision-making (importance, optimism, pessimism, etc). They can arrive at different conclusions even when all given the same data.

I'm sure it's a little of both but I imagine primarily it's #2.

 

[Andre]

Speaking of coordination problems. How many computers are there in a plane and aren't they coordinated and switched off if they don't agree?

Or was it a space shuttle?

Well take the F-16 flight control computer, which basically translates a desire of the pilot indicated by the stick position into a continuous control signal that does everything to hold that. That sounds very basic but it isn't. Moreover the instability of aircraft requires constant computer steered corrections. It's rather hard, but not impossible to do that manually.

The flight control computer is quadruple. Three branches are actually doing the steering, one is in standbye, that is not connected, but operating.

The steering signal is the average of the three and all three are constantly monitored on deviations. As soon as one branch gives a signal exceeding X from the other two, it is kicked out automatically and the reserve system jumps in. The pilot gets a warning.

Should one of the three remaining branches again produce an signal outside the acceptable range then it is kicked out too and the aircraft continues on the two remaining computer branches. But I haven't heard that this ever has happened.

 

[hamster143]

Aside from the questions about autolanding panic buttons & such, I don't see how this is going to make sense financially. I did some ballpark calculations and it seems to me that losing the copilot would save the company at most $4 per passenger, probably less, for a 3-hour flight.

why is the USAF flying multi million dollar machines JSF, F22 etc with just one pilot, if two was safer?

Could be because weight and space in those machines are at a premium. Just the second pilot, his suit and his ejection seat are going to weigh around 200 kg. In a plane that weighs 10-12 tons, that's significant.

And those crash very often anyway for unrelated reasons, so the reduction in risk due to having a second pilot would be negligible. If commercial aircraft crashed as frequently as, say, F-16's, nobody would fly them.

 

[Andre]

Aside from the questions about autolanding panic buttons & such, I don't see how this is going to make sense financially. I did some ballpark calculations and it seems to me that losing the copilot would save the company at most $4 per passenger, probably less, for a 3-hour flight.

which is a significant portion of the earning per passenger. Just check the fares of easyjet.

Anyway, also included training costs, expenses like hotels, etc, in that figure? which is likely a multitude of his salary.

Could be because weight and space in those machines are at a premium. Just the second pilot, his suit and his ejection seat are going to weigh around 200 kg. In a plane that weighs 10-12 tons, that's significant.

And those crash very often anyway for unrelated reasons, so the reduction in risk due to having a second pilot would be negligible. If commercial aircraft crashed as frequently as, say, F-16's, nobody would fly them.

Several fighters are standard in double seat like the F-4 Phantom, F-14 Tomcat, European Tornado albeit that the second crew member is weapon system operator (WSO), but it still would reduce the workload of the pilot, whereas the WSO is capable of handing the aircraft. The question is if that is reflected in the safety records of these aircraft. Are they safer being operated by two? Don't hold your breath.

 

[hamster143]

which is a significant portion of the earning per passenger.

$4 out of the ticket price of $300 is a minuscule reduction in operating costs that can be easily canceled out by any number of things, like higher insurance costs or passengers' slight aversion to fly on single-pilot aircraft.

The question is if that is reflected in the safety records of these aircraft. Are they safer being operated by two? Don't hold your breath.

That's because their safety is poor for reasons that are unrelated to the number of pilots.

Commercial aircraft crash, on average, once per 3 million flight hours. Suppose that losing the copilot quadruples the risk (not an unreasonable assumption, given that general aviation aircraft, usually single-piloted, tend to crash 50 times more frequently than big airliners.) That's one additional incident per one million flight hours.

Suppose that the differential incident rate is the same for fighter jets (one crash that can be prevented by having a second pilot, per million flight hours).

But, since the cumulative crash rate for USAF F-16's since their introduction is on the order of 35 per million flight hours, reducing that by 1 would make no visible difference.

 

[Andre]

$4 out of the ticket price of $300 is a minuscule reduction in operating costs that can be easily canceled out by any number of things, like higher insurance costs or passengers' slight aversion to fly on single-pilot aircraft.

I see I can get from Belfast to Nice for 25 pounds

Suppose that losing the copilot quadruples the risk (not an unreasonable assumption, given that general aviation aircraft, usually single-piloted, tend to crash 50 times more frequently than big airliners.)

Wouldn't that be a bit affriming the consequent fallacy. Couldn't it be that the pilot training, logistics, supportm etc, are on a completely different standard?

Furthermore, repeating once more, how many mishaps in multi crew aircraft would not have occurred with a single crew, because the root cause was failing crew co-ordination?

 

[f95toli]

$4 out of the ticket price of $300 is a minuscule reduction in operating costs that can be easily canceled out by any number of things, like higher insurance costs or passengers' slight aversion to fly on single-pilot aircraft.

As Andre has already pointed out: The "typical" ticket price for a Ryan Air flight is nowhere near $300, last time I flew with them I paid something like £70 for a return flight between London(UK) and Haugesund (Norway). $4 per passenger IS significant for them

Btw, Pilots working for Ryan Air are actually quite well paid (as opposed to the cabin crew, they make much less than the cabin crews working for regular airlines)