Can Aircraft Become *Too* Automated?

[Gym_Class_Hero]

I recently read something that the FAA thinks that the 787 might be threatened by hackers, and it got me thinking, can aircraft become too automated? Look at an aircraft today like the 777, onboard computers can control everything from takeoff to landing, all the pilot has to do is taxi, startup/shut down, and talk to ATC. Aircraft manufactors seem to think that if it has a computer, it is automatically good. Air France 296 pretty much disproved this.

If you ask me, relying too much on computers will end up hurting your flying skills. Back in the day of 727s and DC-10s, a pilot actually had to *fly* the plane. Does anyone else agree?

 

[FlyenRockets]

No.

 

[roadkill]

Ok with me as long as
those computers don't use
Windows Vista....

 

[Bravo52]

Believe it or not, this is a social issue, not technological. There is nothing stopping technology today from transporting people/cargo/weapons from flying totally automated from point to point. That includes everything from start, taxi, t/o and landing. A lot of companies and agencies have demonstrated this over and over with a tremendous degree of reliability.

What hasn't kept up is the ability of government to integrate this technology in the current system. The Air Force (and others) are having a hell of a time with the FAA in letting us fly UASs in the NAS (National Air Space). We talk reliability until we are blue in the face, however, they don't want to give up the control. Granted "control" is used loosely here. It is almost as if safety was not the issue......

Too automated. No. Too late for that. Once people get over the idea of traveling between two points with no pilot at the controls, then it will be a daily occurrence. However, the first step will be to reduce the size of the crew to one pilot (who monitors the computer)........and everyone knows that won't be easy. Too many jobs at stake.

Besides, SkyNet hasn't been formed yet....

 

[user 13]

YES!

Aircraft are already too automated. You need an organic computer in the loop.

No civil passenger aircraft can do auto takeoff. I'm not saying the technology doesn't exist, It's just not done. Computers are great for lots of things, and add great backup. Just don't take the TWO pilots away. NO COMPUTER CAN MATCH AN EXPERIENCED, WELL TRAINED PILOT!

I fly a 737 for a living. It is a very automated aircraft. I have many levels of automation to choose from. Each level has benefits and drawbacks. It's up to me to choose the best for safety. The limitations on automation are huge.

The Auto land on the 737 has very restrictive wind limitations. You could not operate using it for every landing. It does not have the capability.

A computer can not look out the window and make judgment calls on what the weather looks like, how the spacing looks between aircraft, what the wind is doing on approach vs what it looks like on the ground. Can not find wake turbulence, can not cope with Icing.

A computer can not make diversion decisions based on Passenger needs. Can not make a go around decision based on an animal on the runway.


Truth be told, the most useful cues I have on landing is the windsock!

As for UAV's. Keep them out of my airspace. It is crowded enough now, I don't need some pilot less aircraft, controlled by HAL, blundering around the airspace doing God only knows what, not talking to or following the ATC controllers (The 2nd best safety feature in the US).

Yesterday while on a dogleg to final, at 5000ft we were cleared to intercept the localizer. I armed the VOR/LOC on the autopilot. About 3 miles from the localizer course, the aircraft thought it had captured the localizer and started to turn right on the auto pilot. Both myself and the First officer made a derogatory comment about the auto pilot, turned it off and I flew the aircraft, making it do what I needed it to do.


I fly a $60 million aircraft, that has 2 autopilots,2 flight directors, 2 FMC's, 2 INS, 2 GPS, full electronic displays, a heads up display, EGPWS, TCAS, etc. But by far the best safety feature on the aircraft is the 2 well trained, experienced pilots. It will take a LONG time to build a computer that can equal 2 experienced pilots.

 

[dr wogz]

Autromated yes, Monitored no.

Even UAVs have a trained pilot at the helm. Ok, not "in" the aircraft, but in a box somewhere, controlling it, monitoring it, essencially flying it..

The systems help the pilot, not replace the pilots.

Many universities have proven that a self guided / self driven car is a hard thing to accomplish.. It seems they all have some sort of trouble.. This I feel is still in its infancy.

 

[troj]

There's a direct correlation between this and what NASA went through with the manned space program.

Pick up a copy of Digital Apollo sometime. The astronauts were strongly against automation in the Apollo craft -- the infamous "spam in a can" argument. The reality was, the computers were good, darned good, especially when you consider the state of the art at the time they were built.

-Kevin

 

[bobkrech]

YES!

Aircraft are already too automated. You need an organic computer in the loop.

No civil passenger aircraft can do auto takeoff. I'm not saying the technology doesn't exist, It's just not done. Computers are great for lots of things, and add great backup. Just don't take the TWO pilots away. NO COMPUTER CAN MATCH AN EXPERIENCED, WELL TRAINED PILOT! ....

I fly a $60 million aircraft, that has 2 autopilots,2 flight directors, 2 FMC's, 2 INS, 2 GPS, full electronic displays, a heads up display, EGPWS, TCAS, etc. But by far the best safety feature on the aircraft is the 2 well trained, experienced pilots. It will take a LONG time to build a computer that can equal 2 experienced pilots.

Well said Mark.

I still cringe every time I see the AIRBUS crash video in Toulouse when the Chief test pilot and Airbus execs were demonstrating touch and goes to a potential customer and the flight control computer wouldn't let the airplane climb out. Unfortunately everyone was killed when the plane crashed in the woods.

Bob

 

[flygrimm]

posted by mach7

'Yesterday while on a dogleg to final, at 5000ft we were cleared to intercept the localizer. I armed the VOR/LOC on the autopilot. About 3 miles from the localizer course, the aircraft thought it had captured the localizer and started to turn right on the auto pilot. Both myself and the First officer made a derogatory comment about the auto pilot, turned it off and I flew the aircraft, making it do what I needed it to do.'

One of my favorite sayings comes from one of our ground instructors at my work is 'upgrade to manual'.

like the 737 the Lear 45 is highly automated and sometimes has a mind of it's own. As a pilot I can't see a pilotless commercial airplane anytime soon. These things after all are programmed by humans. And we make the same comments about our autopilot as well.

Stuart

 

[Bravo52]

It is very rare that a plane crash isn't caused by "Pilot error".....that includes judgment as well as stick and rudder ability.

All of the things said so far are true. But, the reality is a computer can and does do all of those things. They will only get better and cheaper. I've been a pilot for over 28 years and agree with all of the emotion very well laid out by Mach7, but the days of no/low automation are numbered.

Here's a pretty good story. A pilot from airline X was flying a sorite across the Atlantic. When he got on the ground the check pilot handed him a pink sheet for his checkride. "But you weren't even on the jet" said the pilot. "I know, but we were monitoring your level offs, throttle positions and your needle deflection on approach. According to the flight data recorder, you were a half a dot low at intercept and according to the radar altimeter you flared too high."


Hmmmmm.......has that happened to you yet? I don't know what the airlines are doing with the data, but it does get collected.

This is an interesting discussion! But Bob, bad choice on which crash to cite!

According to the pilot of that one, "Captain Asseline flew the aircraft manually. He had been instructed by Air France to overfly the airfield at 100 ft above ground. When he increased throttle to level off at 100 ft, the engines did not respond. So after some seconds he got worried and thought there was something like a short-circuit in the completely computerized throttle control. So he pulled the throttle back all the way and forth again. By that time the aircraft had touched the trees. "

More on that here:

https://www.airdisaster.com/investiga...96/af296.shtml

Look, I'm no expert, but I believe the future is "Pilot Optional". At least in the military. And where the military goes, so goes the public sector.....at least eventually.

 

[Viperfixr]

Funny, just last Saturday I was sitting around a fire pit sharing some scotch and cigars with a couple senior AF officers, talking about the F-35 vs. F-22 budget debate. One officer thought the real threat wasn't that at all, but rather the Lightning II being the last manned fighter in AF history!

While I wouldn't hesitate to have an X-45/47 flying in a battlespace unmanned (not dropping near friendlies, but deep threats instead), I would not be excited to fly in an airliner that was lacking a real, human pilot up front.

 

[user 13]

Bravo52,

You are correct. The future is pilot less, just not the foreseeable future.

Most accidents are pilot error, but not all. Another question to ask is how many
accidents are AVOIDED by pilot interaction.

How many airline flights are flown each year with how many crashes? What kind of machine or system can come close to that kind of reliability?

 

[Bravo52]

Another question to ask is how many accidents are AVOIDED by pilot interaction.

Excellent point! However, some might say; "not enough".

I don't expect to see pilotless airliners in my life time. I think you will see the regs change where we have fewer pilots in the cockpit. You know very well you could fly the 737 single pilot. Much less with a better automated system. Admit it, you are thinking about one or two of those copilots you flew with and thought "I'll call him....wedge, the simplist tool known to man".

The first rule to fall will be the demarcation between small and large aircraft (ie the 12,500 lb rule). The King Air 300, 350, CJs, Beech Premier a lot of Cessna jets and and many more are all single pilot.

Granted, those don't carry a lot of little warm bodies, but they do fall into the same catagory as the 737 (sort of). Oh, and the 737 might be the first of the really big private jets to be single pilot. I know I sure wanted two pilots in the 737-200s........

 

[Gym_Class_Hero]

Doesn't John Travolta fly his 707 by himself?

 

[les]

On many of the new arcraft today (such as the F35), the stick has no "direct" control over the surfaces. The stick sends signals to a computer system that then (based on various other parameters such as air speed, attitude, etc) moves the surfaces via actuators.

Basically, even these aircraft are at the mercy of the computer system. If the computers "disagree" with the pilot or fail (although they have redundancy) the pilot has no means to control the aircraft.

The autopilot is just another software program within the computer system that provides the basic functions like maintianing heading or altitude to the more complex auto-take off or landing.

 

[Peartree]

Remember that the Army wants driver-less trucks because robots don't fall asleep and no one cries if they get blown up. The Air Force wants pilot-less fighters because no one cries when they get blown up (except for the taxpayers) and because pilots can't pull 10 g's and airframes can.

 

[Fred22]

Remember that the Army wants driver-less trucks because robots don't fall asleep and no one cries if they get blown up. The Air Force wants pilot-less fighters because no one cries when they get blown up (except for the taxpayers) and because pilots can't pull 10 g's and airframes can.

I think UAVs are the wave of the future but not right now. If the aircraft has it data sent to the ground what prvents some nut from telling it to fly into the ocean or the ground? Pilots are indispensible right now I think. In as regards to automated trucks I think what they want is 0ne truck in a certain number to be manned and all the rest just play follow the leader so if he gets it what happens then ?
Apparently robby the robot has not been quite up to snuff in Iraq
https://www.popularmechanics.com/blogs/technology_news/4258963.html
I really like the story of how one was disabled with a semipermeable wool barrier or in other words someone threw a blanket over it in testing renderring it useless
Cheers
fred

 

[user 13]

Doesn't John Travolta fly his 707 by himself?

If you believe him he does. Truth be told he always has a qualified instructor with him. I have flown with guys who tried to train him. Scary.

Too much money, too little brains.

 

[georgegassaway]

Pick up a copy of Digital Apollo sometime. The astronauts were strongly against automation in the Apollo craft -- the infamous "spam in a can" argument. The reality was, the computers were good, darned good, especially when you consider the state of the art at the time they were built.

In the days of Mercury, they wanted to PILOT the entire launch from liftoff to orbit. Not let the automated guidance system do it. That was just nuts, I do not care how good a pilot is, they cannot fly a rocket launch vehicle as smoothly and accurately (and as efficiently) as a rocket's guidance system can.

Look at how close Space Ship One came to disaster on that flight tat got into an out of control roll situation. They gambled on using a pilot to do it manually to save money and time over an automated system, and they almost paid dearly for it (not as dearly as the pilot would have). But even at that, they did not have anywhere to "go", such as achieve an orbit, or even worse an orbit to try to dock. If that flight's objective had been to fly through a projected 1/2 mile diameter "hole" in the sky at 100km up, they'd have missed widely. And if that flight had carried paying passengers and the goal was not to scare the *** out of them by the vehicle being out of control for awhile, they'd have missed that widely too.

All that said, for most situations such as passenger airliners, it is a good thing for the human pilots to be involved with flying the aircraft for takeoff and landing, and other critical parts. For space, it is best for the computers to do that for launch and re-entry. For docking, manned seems best, at least when there is truly a human onboard the vehicle doing the docking. I do not know how the Russian "Progress" resupply ships do it now, if it is auto or remotely flown. I do know of an accident with MIR where a cosmonaut inside of MIR was remotely flying the Progress and things went terribly wrong resulting in a collision. I do not recall but think that Progress has done automated dockings before, so the only collision was the one involving human remote control.

Lunar Landings, I think those are best as "semi-Auto", where the computer is capable of doing the landing all by itself, but a pilot can take over and land manually if they feel something is not quite right or see they are about to land in a bad place (such as when Armstrong actually flew the Eagle LM horizontally a long way from the rock-infested crater the computer was bringing them over).

For the shuttle orbiter, it is capable of flying itself all the way to landing, except the landing gear switch has to be thrown manually. But the pilots do not want to lose out the opportunity to land the orbiter manually. So at about 20-30,000 feet or so, they go from auto to manual, and fly the rest manual. Even if you ignore that the orbiter is Fly By Wire, even that last "manual" part is not exactly "manual" like one would fly a Piper cub, or even a 747, since as a big heavy glider, it has to fly along a pretty narrow flight corridor, both vertically and horizontally, so there are computer-generated path lines in the Heads Up Display showing the Pilot/Commander exactly where to steer horizontally and vertically and they also of course must maintain the proper range of velocity as well. There is also a computer generated "energy management" indicator, balancing out the altitude, velocity, and projected distance to the runway.

So if one thinks that by flying the orbiter "manually" means that it is being done in case the computer might be wrong, forget that. Because if the computer would be wrong about an automated landing, then the HUD data and energy management data would also be wrong, and the pilots will be lead into flying the orbiter inaccurately and never make the runway. Sort of like letting a flawed car GPS system "make you" manually drive west into a river instead of using the bridge 1000 feet to the north of you. But as a car driver you could see at the last moment there is no bridge there, and drive to the actual bridge. In the shuttle, by the time the pilots realized the HUD and energy management data was wrong it would be too late.

- George Gassaway

 

[user 13]

In X-Plane I have "flown" the X-20 manually from the B-52 drop at 50000 ft. It can be done but it is very hard. I have also flown the Shuttle in X-Plane and in the sims at various museums, It is much easier with the HUD and computer guidance. With both I like to fly the final approach (last mile) visually. The computer cues just don't seem to be correct to me.

Remember Apollo 13, they manually flew the reentry.

 

[georgegassaway]

In X-Plane I have "flown" the X-20 manually from the B-52 drop at 50000 ft. It can be done but it is very hard. I have also flown the Shuttle in X-Plane and in the sims at various museums, It is much easier with the HUD and computer guidance.

I have done the Shuttle in X-Plane too. The only part that is sort of practical to do without the HUD and guidance info is the really short one that is little more than a straight-in approach. And the monster is to fly the whole re-entry manually, something the astronauts have never done (they let the computers fly it till at least well after re-entry). Doing the full re-entry manually, the slightest error and you may get too low in the atmosphere, then you cannot climb back up because the air is so dense that you slow more and more, then get out of control and game over. Or fly too high for the given velocity and stall out due to the thin-ness of the atmosphere, then either lose control or in trying to recover, to get too deep into the atmosphere for that velocity. I think I have made it thru a few times, but it is both so boring for a long stretch, when you may be getting into deep trouble that by the time you do realize a problem and it is not boring.... it is probably too late. So after doing that a few times, I went back to the option that runs the last 5 minutes or so, not the whole re-entry.

Remember Apollo 13, they manually flew the reentry.

Apollos did not so much "fly" reentry, they were "falling, with style".

And still, it was whole lot easier to "nudge" a ballistic path to land a little left or farther out, with no real consequences for missing by miles since there was the whole Ocean to land in, and three parachutes to open. Way way more complex deal with the type of lifting re-entry of an orbiter where it is skimming the atmosphere for many many times farther than Apollos did. And the orbiter only very gradually gets lower into the atmosphere, until the last few minutes when it is past re-entry and is deep enough into the atmosphere to start slowing down a lot more dramatically (the complex dynamics of the orbiter lifting re-entry process are almost mind-blowing, even ignoring the heat and the danger). And near the end, to fly those manuevers to do the turns just right, not too high, not too low, and with the correct amotn of enery left, to land withni a few feet left or right of the runway and not too far down the runway (or worst of all, short of the runway),with something that glides as badly as that does (badly in relative terms anyway. For what it is, it does well). We ain't gonna see Orion landing on or near the KSC runway unless something goes horribly wrong.

- George Gassaway

 

[user 13]

Falling with style!

I like that, very nice words-man-ship.

The whole shuttle thing is truly amazing. A no power approach, starting from 1000's of miles away, to land in the first 3000 ft of a 15000 ft runway. With so many critical point along the way.

But back to the original topic. Sure the shuttle can fly and land on automatics, but again the weather requirements are very restrictive. Particularly the wind.

 

[DexterLB]

Ok with me as long as
those computers don't use
Windows Vista....

LoLoLoLoLoL

Very good point!

 

[georgegassaway]

Falling with style!

I like that, very nice words-man-ship.

It comes from the movie “Toy Story”. I will borrow a description from the web:

“Two of the main characters are Woody, a 1950s style cowboy action figure, and Buzz Lightyear, a spanking new spaceman action figure who comes into the playroom as a birthday gift to Andy, the little boy to whom the toys belong. Buzz has wings, and Buzz believes he can fly " to infinity, and beyond!". Woody is the naysayer. "You're a toy! Toys can't fly." So what does Buzz do. He dives into the emptiness, and through a series of 'coincidences', flies to another part of the room. Woody, staid, steady Woody, says That's not flying. That's falling...with style.”

YouTube video (not great quality):
https://www.youtube.com/watch?v=4gGc1j5A2_A

But saying Apollo’s CM was “Falling with style”, yeah, I’ll take credit for that...

But back to the original topic. Sure the shuttle can fly and land on automatics, but again the weather requirements are very restrictive. Particularly the wind.

Crosswind, yes. Unlike airliners, they cannot even land in rain, or it will erode the tiles horribly, almost like sandblasting but with water drops (well, I think they could at least land safely, but at this point of the remaining schedule it would be impractical to replace the tiles to fly again).

The good thing is that they can choose when they come down, so if the weather and wind forecast is reasonable 45 to 50 minutes before the planned landing, they go for the de-orbit burn, and if the forecast is not good they can wave off and perhaps try one orbit later if the forecast was for a temporary problem. If that is bad, then they can try the next day for KSC, or if things are going to be bad at KSC for several days in a row, go for Edwards AFB in California. If crosswind might be an issue for the big primary runway at Edwards, they could also land on one of the “painted” runways on the dry lakebed, once of which would be of an acceptable crosswind vector, unless there is simply a hell of a windstorm going on.

I must admit that I have no been following too closely the most modern unmanned military aircraft regarding how they simply take off and land. I know they started out little more than oversized R/C model airplanes, with a TV camera onboard. But I had thought there have been a couple by now that are 100% automated, capable of taking off and landing all by themselves without a human to do it?

And of course the Russian “Buran” shuttle, on its one and only flight in 1989, was unmanned, landing by itself quite smoothly (I do not know what the winds were like).

Anyway, to really go back to the original question of this thread, about hacking, does it matter then if there are human pilots onboard or not? If someone hacked the shuttle’s computers, or ANY fly-by-wire aircraft, there is nothing that the pilots can do to save the aircraft if there is a virus screwing up the programming to make the elevators move down, or even an unintentional programming error that would have the same effect.

Or that B-2 bomber that crashed on takeoff a year ago. As with the shuttle, it is the epitome of an aircraft that has to use fly by wire, no direct “stick and cables” mechanical controls.A quote from the news:
“The crash was caused by distorted data in the aircraft's flight control computers, according to a report released Thursday. Air Force investigators blame the distortion on moisture in the system, which caused the computers to calculate the wrong airspeed.”

Of course the pilots were controlling the B-2, and the computers were converting their control inputs into the control surface moments to produce the desired flying response. But with the air sensors (I am not sure if those were some advanced angle of attack sensors of more like Pitot tubes, but I think AOA) screwed up, the computers were lacking the accurate feedback info they needed. And there was nothing the pilots could do other than what they did... eject.

Now, take that example of screwed up air data sensors and imagine someone hacking the system of other fly by wire aircraft to simulate bad data. Something that would not be noticeable on the ground. Or even something as classic as the ailerons being reversed. I cannot recall exactly which, but know of some British prototype airliner around 1940’s or 1950’s that crashed on takeoff because the ailerons were installed reversed, and somehow nobody ever realized it in ground testing (I have to admit to having had that happen with my R/C models a couple of times, due to using different models with different set-up requirements using the same transmitter).

- George Gassaway

 

[dwmzmm]

How about Buran's first and only flight? Fully automatic and unmanned. Even
touched down the runway with a good crosswind.....

 

[DMcCauley]

You're never going to see fully automated commercial flights because of one thing. Unions. Just like you don't see the type of automated and robotic assembly systems common with cars overseas at US car manufacturing plants.

 

[georgegassaway]

You're never going to see fully automated commercial flights because of one thing. Unions.

One thing - Passengers.

Most passengers would not fly on an airliner that was not piloted by a human. At least not in the foreseeable future. Maybe after a hundred years of technological develpment, and many decades of established reliable operation by the US Air Force.

- George Gassaway

 

[Viperfixr]

Remember that the Army wants driver-less trucks because robots don't fall asleep and no one cries if they get blown up. The Air Force wants pilot-less fighters because no one cries when they get blown up (except for the taxpayers) and because pilots can't pull 10 g's and airframes can.

Ummm, pilots run the AF, so do *really* think that they want pilotless aircraft? Not really. Not universally. You should hear the criticism of the Global Hawk, especially from the U-2 community--even Intel. I was involved in an X-45 site survey, and the jet seemed incredible in many ways--but it's gone too. Have you read about the pissing contest between senior USAF and USA brass over UAVs?

We'll use true UCAVs (not the Predator or Reaper) at first for going after highly defended, high value, low collateral targets. Then it will grow from there, but it will take TIME. There are no open AF arms for UCAV/UAV acceptance.

 

[user 13]

The autoland systems have strict wind limits all around. Head, cross, and tailwind.

I believe the airbus has 5 primary flight computers. 3 Motorola and 2 Intel chips. They have different operating systems and don't swap data. Just to make it hard to corrupt. If everything goes bad it reverts to a basic system that gives exactly what the pilot commands, no processing. Boeing's system is similar, but is more pilot oriented.

George, Thanks for the source of the line. I saw Toy Story too.

 

[Bravo52]

Ummm, pilots run the AF, so do *really* think that they want pilotless aircraft? Not really. Not universally. You should hear the criticism of the Global Hawk, especially from the U-2 community--even Intel. I was involved in an X-45 site survey, and the jet seemed incredible in many ways--but it's gone too. Have you read about the pissing contest between senior USAF and USA brass over UAVs?

We'll use true UCAVs (not the Predator or Reaper) at first for going after highly defended, high value, low collateral targets. Then it will grow from there, but it will take TIME. There are no open AF arms for UCAV/UAV acceptance.

That's too broad a generalization. Too much emotion. I can tell you that there is a lot of interest from the Sr. leaders for unmanned systems. The focus of future requirements will be man-optional......err....person-optional. MONEY drives things much more than TIME. RISK is evaluated and considered more than TIME.

Just like the stealth technology in the F117, it's served its purpose and we've moved on from there. The next technological advance is in electrons. The technology of the F117 got too expensive to retain based on the advantage it provided. We found out that we can do the same thing with a Reaper and or an F-16 at a much lower cost with the same/less risk.

The pissing contest between the AF and Army over UAS had nothing to do with capability. Has to do with control.....oh, and money. Army wants centralized control, centralized execution. The AF does not.

I would also argue that the problem the "AF" has with UASs has more to do with pilots wondering what they'd be doing vs. can a UAS do it.......think tribal.