One in Two People Say They Won’t Fly in Fully Automated Aircraft

In some circumstances, pilots do save lives after a mechanical issue or a bird strike. However, it is estimated that 80 percent of accidents are caused by pilot error.

http://www.bbc.com/travel/story/20130521-how-human-error-can-cause-a-plane-crash

If only 20 percent of accidents are caused by mechanical/weather/external issues, we might actually come out ahead even if we assume that a human pilot is always successful in that 20 percent situation and autopilot always fails. In reality not every pilot is a Sully and I would guess that autopilot might not always fail in an emergency situation. Just a few weeks back an Air Canada flight almost landed on two planes because the pilots aimed for the taxiway and not the runway.
Need to be careful when interpreting those stats. 80% of accidents doesn't tell us how many accidents were averted by human pilots. If there are 100,000 flights per day and 1% of them have something go wrong but only 1% of *those* flights result in an accident, we're talking about 1,000 potential accidents but only 8 accidents caused by human error. That means 2 accidents weren't caused by human error but 990 potential accidents were avoided presumably by corrective action by human pilots. The 100K flights per day is accurate, you'll have to search the percentage of potential accidents and actual accidents and plug the numbers in to assess the extent of the issue, but that's how you need to run the numbers in order to evaluate human vs. automated flight safety and not just how many accidents were caused by each.

You can also get rid of turbulence by building the seats onto big shocks, like the magnetorheological style. That would be nice =P
Big shocks? Come on, when was the last time you flew? In a couple years we'll be lucky if the airlines aren't stacking passengers nut to butt.
 
I'm not saying that I won't ever fly on a fully automated aircraft, but I will definitely give the tech a few years to mature. And do you guys really think that a totally automatic flight system could pull off what Capt. Chesley Sullenberger did? I don't. But, on the flip side, a fully automated system would cut back a lot on aircrew error.

Basically, all you really need to do is tell the AI to "find someplace relatively flat, preferably an airport if possible", and "keep the plane horizontal", and you've accomplished basically the same feat as Sully.

Lets face it, what happened was essentially:

1: Engines lose power immediately after takeoff; plane becomes glorified glider
2: Understanding there isn't enough altitude + airspeed to turn around and make the airport again, requiring immediate landing in a relatively flat area (given NYC, this mandated a water landing)
3: Point plane at designated landing point (Hudson river, which is a pretty big flat target)
4: Glide into water at "reasonable" (not-too-fast on landing, not-too-slow you stall)
5: Thank the Federal regulation that mandates passenger planes be buoyant for at least 90 seconds after a water landing.

Honestly, what Sully accomplished is literally the easiest case for AI to handle. Loss-of-Engine-Power cases are relatively "simple", since you can just treat the plane as a glider at that point. Other failures [loss of control surfaces] are much harder to deal with by comparison, since you need to factor in aircraft control into the decision making process.
 
Wow is this thread full talking without knowledge...I wont address everythin (there isnt time or space for it). But heres the first ones I saw

Exactly. Modern jets are capable of takeoff and landing on their own

Not quite true. Many are not equipped for it and fewer airports are equipped for it.

A rare exception to a general rule.

Less rare than you might think.

http://www.nytimes.com/2011/07/30/world/europe/air-france-flight-447-crash-report-july-2011.html
https://www.washingtonpost.com/loca...6ffb874a18d_story.html?utm_term=.8cab6733c231

Humans aren't much use when autopilot fails since it could be ages since they last did anything manually. These are pilots with plenty of experience but they surely don't train they way they did when they came out of their military careers in commercial flight.

Maybe they should just have drone pilots whose jobs are to train/fly constantly (in low risk conditions such as mixed simulator and remote control of actual unmanned training aircraft) and then take over when an actual autopilot failure occurs. With ipad toting pilots, it's like homer simpson running the nuclear plant but IRL.

That is because people get lazy. The pilots in these cases should have said "hey I am not PROFICIENT even though I am current" and not flown. I have turned down paying flights due to what I felt was insufficient proficiency despite being legally current or I just went out and flew some practice to get proficient.

We will never know how many pilots could have pulled that off, I'm going to guess very few. http://www.airspacemag.com/as-interview/aamps-interview-sullys-tale-53584029/

Most pilots, who are current in their aircraft, could have pulled that off. We are trained for it. I hate to critize the guy but some of his statements were clearly slanted to make it sound a bit more exotic. He talks of energy management being no big deal for small planes...he should know better from his days flying in the military or hell flying PERIOD. Its all about energy management not matter what you fly.
 
I wonder if they realize that modern airliners are *ALREADY* fully automated?

A modern 787/A380 can go from runway to runway without the pilot touching a single control.
 
Most pilots, who are current in their aircraft, could have pulled that off. We are trained for it. I hate to critize the guy but some of his statements were clearly slanted to make it sound a bit more exotic. He talks of energy management being no big deal for small planes...he should know better from his days flying in the military or hell flying PERIOD. Its all about energy management not matter what you fly.

And Sully is an asshole. Have a couple friends that worked at the same airline. Nobody could stand him. But he was a "hero," so publicly they had to compliment him at every opportunity.
 
its all fun and games until someone puts a few cleverly placed strips of tape on the runway,
 
  • Like
Reactions: N4CR
like this
Yeah they won't. Like there is shit they can do about it. When it comes down to it you either fly with it or you don't fly at all.
 
its all fun and games until someone puts a few cleverly placed strips of tape on the runway,
kek.png
 
Well, this is kind of sad because being a pilot is such a dream job for lots of people. Flying is a passion for many and being able to spend your life flying around the world is the ideal life for people with that passion.

Considering costs, flying will be limited to small aircraft now and no human will ever again know the feeling of flying something like a passenger jet... except maybe in the military.

I guess the cockpit will now be the ultra-first class.
Truthfully flying big airliners is about as interesting as being a passenger on one. At 35,000 ft the world is mostly flat in appearance, and pilots easily tune out to attend to other things rather than enjoy the view, of which is not particularly glorious looking through tiny cockpit windows.

Real flying is done in GA aircraft that you own/rent and joyride, where you can see the details of the land below, flying among the clouds or over the hills. You'll always have this option. Think of it like riding a motorcycle through the back roads vs riding a bus through the city. Is the bus driver really sad when he loses his job?
 
  • Like
Reactions: N4CR
like this
I'm not saying that I won't ever fly on a fully automated aircraft, but I will definitely give the tech a few years to mature. And do you guys really think that a totally automatic flight system could pull off what Capt. Chesley Sullenberger did? I don't. But, on the flip side, a fully automated system would cut back a lot on aircrew error.
Thing is, whatever mishaps occur on a daily basis during flight tend not to suffer severe consequences. Planes arent crashing daily. Airline pilots are pretty damn good at their jobs, so when coupled with the human element to correct mistakes computers werent programmed for I kinda dont see the point. I have about 40 hours under my belt flying single engine Cessna's, and sometimes my landings leave a bit to be desired. Could a computer grease it every time where I cant? Yeah probably, but it's not like I ever almost crashed a plane. What few imperfections I have in my landing outweigh what happens if I had an engine failure and had to land on a golf course.

Ultimately I just dont see what value a FULLY automated flight system would have. They say 70% of flight errors are caused by humans, but what exactly are those errors? Is it anything to even be concerned with? Minor error in driving a car can easily cause an accident. Surprisingly minor errors in flight are easily correctable.
 
Last edited:
Yes, some of us like to get paid to do what we love...
What I'm saying is being an airline pilot isnt that glorious and you can still do what you love and get paid for it, you'll just be doing it differently. Ask any pilot which aspect of flying he enjoys more, cruising around in his 2 seater bubble canopy or sitting behind a control deck with windows that give just enough visibility to see the runway and land.
 
Airbus' history is rife with accidents caused by the engineers not foreseeing what could happen.

You can't program your way out of a situation...unless you accounted for EVERY variable beforehand. If you're that prescient, you would've already cornered the stock market and perhaps picked up a powerball or two.
 
I wonder if they realize that modern airliners are *ALREADY* fully automated?

A modern 787/A380 can go from runway to runway without the pilot touching a single control.

Airbus' history is rife with accidents caused by the engineers not foreseeing what could happen.

You can't program your way out of a situation...unless you accounted for EVERY variable beforehand. If you're that prescient, you would've already cornered the stock market and perhaps picked up a powerball or two.

Have you heard about Air France?
 
Ultimately I just dont see what value a FULLY automated flight system would have. They say 70% of flight errors are caused by humans, but what exactly are those errors? Is it anything to even be concerned with? Minor error in driving a car can easily cause an accident. Surprisingly minor errors in flight are easily correctable.

80% and it includes everything where humans have input, from manufacturing parts, to fitting/replacing them, to pushing buttons as a pilot.

Removing the pilot is merely a cost saving measure. They can save maybe 100 million a year in wages alone in some of the larger airlines... now you see why self driving is coming regardless? Trucks, taxis, trains, busses, planes, ships.
All we really need is maintenance people and a few skilled drivers/pilots/riders for applications too challenging (so far) for AI. Crop dusting, rescue work etc etc

We should really be saying, okay unmanned flights, IF you give us a cheaper flight...
 
Airbus' history is rife with accidents caused by the engineers not foreseeing what could happen.

You can't program your way out of a situation...unless you accounted for EVERY variable beforehand. If you're that prescient, you would've already cornered the stock market and perhaps picked up a powerball or two.
It's a risk assessment. Does the risk of having an automated system which will occasionally fail outweigh the risk of having a human system that will frequently fail? In the case of automated driving I agree. Everyone is concerned about cars that decide to kill the driver instead of that drunk pedestrian walking into traffic, or little timmy getting splattered on his bike because Tesla didnt want to send the car into a tree. Thing is those situations will account for probably 0.0001% of incidents whereas all of the other accident avoidance will be well worth it.

However we dont have these problems in air traffic. We've been flying by hand for decades with hardly any errors at all, namely because there arent that many planes in the air in relation to each other. Now if future air travel looked like something out of back to the future II or the 5th element, then yeah automated air travel would make more sense. But current day air travel is prohibitively expensive and prevents this kind of volume from occurring. The issue solves itself. At any given moment most planes are several miles apart from each other in all phases of flight so there's very little risk of accident, even upon landing.
 
The issue solves itself. At any given moment most planes are several miles apart from each other in all phases of flight so there's very little risk of accident, even upon landing.

Four words: Reduced Vertical Separation Minimums...
 
The average human driver makes more errors compared to a self driving current car. If all cars would be networked (aware of eachothers locations, speed and intentions) the current rate of accidents could be dropped to a small fraction. Only equipment failure would then lead to accidents.

That is true. But if I am driving across the bay bridge and my gps says to make an immediate right, I am just going to laugh and keep going. A self driving car will actually turn right and take you for a swim.
 
Can these people in the thread passing off supposed facts please tell me what they do for a living?

Reason I ask is that I wouldn't take any supposed facts to do with automated vechiles from someone working on banking software, etc.

Thanks
 
  • Like
Reactions: PaulP
like this
Can these people in the thread passing off supposed facts please tell me what they do for a living?

Reason I ask is that I wouldn't take any supposed facts to do with automated vechiles from someone working on banking software, etc.

Thanks

I am a commercial pilot and a certified flight instructor.
 
Aircrafts are flown by computers about 98% of the way as it is already. It's got to a point as where pilots are struggling to have enough 'hands on stick' time to keep their flight verifications despite working full hours on commercial flights. Also no modern jetliner does anything without a computer anymore. The pilots stick inputs are just suggestions to the flight computer on what to do. The newest planes will even stop the pilot from doing excessive movements that would endanger the plane for example.

Most accidents happen because the pilots disengage the automation and give wrong inputs in confusion. Of course sometimes a problem may disengage the autopilot by design but that should obviously never happen in a fully autonomous airplane lol.

You know, the logical side of me totally gets this. After being in the software industry for 20 years, my emotional side still trusts human reaction and judgement over a computer. I think I just have to get used to the idea and know that airplanes are flown mostly by computer already. Thanks for the info.
 
Its like calling the pilot of the ditched plane a hero....NO he just saved his own ass and the others were there to see it close up.
Without a pilot you are on your own and if something goes wrong you are another statistic....the airline told him to return to the airport so fly thru the city to get there
Just a way to save more money. How about illegal aliens with some flight training for $4 an hour....maybe thats their other plan.

1 out of 2....I got on a gov super computer and after 4 days of constant grinding it seems that it comes close to 50%....what moron wrote this
 
Flying in a plane is unnatural enough already heck if care is a computer or not. Feet are made for the ground... Oh well.
 
That is true. But if I am driving across the bay bridge and my gps says to make an immediate right, I am just going to laugh and keep going. A self driving car will actually turn right and take you for a swim.

The car wouldn't be guided just by its GPS. It would know that there's a bridge rail and between it and a large body of water and not turn either.
 
You know, the logical side of me totally gets this. After being in the software industry for 20 years, my emotional side still trusts human reaction and judgement over a computer. I think I just have to get used to the idea and know that airplanes are flown mostly by computer already. Thanks for the info.

Except hes wrong. I used to think this myself but I learned that was mostly a myth. http://www.cnn.com/travel/article/autopilot-airlines/index.html
 
Except hes wrong. I used to think this myself but I learned that was mostly a myth. http://www.cnn.com/travel/article/autopilot-airlines/index.html

Wrong. That was just a pilot saying they're still needed in the cockpit. The article actually verified everything I said right from flying most of the part automated (in rare cases even landing on autopilot) and to the flight computer blocking the pilot from doing illegal moves (on Hudson landing).

Airbuses have a similar flight law limitations like the F35 for example has. The plane can pull all sorts of things up to midair breakage but the computer limits the pilot from doing stupidities.

If you watch aircraft disaster investigation stories you'll find that the majority of disasters are either directly caused by pilot error or a mechanical problem combined with pilot error. Very rarely a single problem alone leads to disaster. It's always a series of events and bad decisions.

One of the most common causes to crashes is a strange warning signal(s) which leads to the crew doing a check list (as mandated by the rules). Many aircraft have crashed because the crew has been so busy with a problem or series of problems that they actually forgot to fly the plane until they ran it out of fuel or lost control.
 
Last edited:
Basically, all you really need to do is tell the AI to "find someplace relatively flat, preferably an airport if possible", and "keep the plane horizontal", and you've accomplished basically the same feat as Sully.

Lets face it, what happened was essentially:

1: Engines lose power immediately after takeoff; plane becomes glorified glider
2: Understanding there isn't enough altitude + airspeed to turn around and make the airport again, requiring immediate landing in a relatively flat area (given NYC, this mandated a water landing)
3: Point plane at designated landing point (Hudson river, which is a pretty big flat target)
4: Glide into water at "reasonable" (not-too-fast on landing, not-too-slow you stall)
5: Thank the Federal regulation that mandates passenger planes be buoyant for at least 90 seconds after a water landing.

Honestly, what Sully accomplished is literally the easiest case for AI to handle. Loss-of-Engine-Power cases are relatively "simple", since you can just treat the plane as a glider at that point. Other failures [loss of control surfaces] are much harder to deal with by comparison, since you need to factor in aircraft control into the decision making process.
The problem with your theory is that one of the engine sensors was telling the automation that it was still running, when in fact it was not. If a computer would have been flying that plane, it would have believed the sensor and tried to return to the airport, crashing before it got there. Sully knew the engine was dead and instead concentrated on getting the aircraft down safely. He also knew to skip ahead in the emergency checklist and get the APU running ASAP so that he had power to continue flying the plane. Computers just follow the rules as they are coded - every time, with no exceptions. It's one of the things that makes them such useful tools. But also only tools, and not in charge.
 
Wrong. That was just a pilot saying they're still needed in the cockpit. The article actually verified everything I said right from flying most of the part automated (in rare cases even landing on autopilot) and to the flight computer blocking the pilot from doing illegal moves (on Hudson landing).

Airbuses have a similar flight law limitations like the F35 for example has. The plane can pull all sorts of things up to midair breakage but the computer limits the pilot from doing stupidities.

If you watch aircraft disaster investigation stories you'll find that the majority of disasters are either directly caused by pilot error or a mechanical problem combined with pilot error. Very rarely a single problem alone leads to disaster. It's always a series of events and bad decisions.

One of the most common causes to crashes is a strange warning signal(s) which leads to the crew doing a check list (as mandated by the rules). Many aircraft have crashed because the crew has been so busy with a problem or series of problems that they actually forgot to fly the plane until they ran it out of fuel or lost control.

Are you a pilot?
 
I have been developing embedded, real-time software for 35 years. I've worked on Air Traffic Control systems for the last 8 or so (Common ARTS and Micro E-ARTS). Statements like: "95% of passenger flights are fully automated", and "modern jets can almost fly themselves" are very misleading, if not just plain wrong. For example, when using the auto-pilot, FAA rules still require a qualified pilot to be sitting in one of the flight seats and be prepared to take control at any time. There are many good reasons for this. If the TCAS warning goes off, the auto-pilot can't follow the directions to avoid the conflict; a human pilot must do this. Or if the EnRoute controller contacts the aircraft (voice via radio) in order to change the assigned altitude or change course for some reason, there must be a pilot there to do it. There are so many things that can go wrong, a human is required in the loop and will be for quite a long time. The FAA is automating more and more functions, but it is slow to roll them out due to the serious safety considerations they must take into account. For example, the DataComm system, which will replace many standard voice exchanges for things like requesting and receiving clearance to take-off and land, has been developed, but it won't be rolled out for several years yet. It takes a long time to test something like this and then train controllers and pilots on how to use it. So good luck to Boeing, but their plans are very long term and I probably won't live long enough to see a fully pilotless passenger aircraft.
 
  • Like
Reactions: kju1
like this
I have been developing embedded, real-time software for 35 years. I've worked on Air Traffic Control systems for the last 8 or so (Common ARTS and Micro E-ARTS). Statements like: "95% of passenger flights are fully automated", and "modern jets can almost fly themselves" are very misleading, if not just plain wrong. For example, when using the auto-pilot, FAA rules still require a qualified pilot to be sitting in one of the flight seats and be prepared to take control at any time. There are many good reasons for this. If the TCAS warning goes off, the auto-pilot can't follow the directions to avoid the conflict; a human pilot must do this. Or if the EnRoute controller contacts the aircraft (voice via radio) in order to change the assigned altitude or change course for some reason, there must be a pilot there to do it. There are so many things that can go wrong, a human is required in the loop and will be for quite a long time. The FAA is automating more and more functions, but it is slow to roll them out due to the serious safety considerations they must take into account. For example, the DataComm system, which will replace many standard voice exchanges for things like requesting and receiving clearance to take-off and land, has been developed, but it won't be rolled out for several years yet. It takes a long time to test something like this and then train controllers and pilots on how to use it. So good luck to Boeing, but their plans are very long term and I probably won't live long enough to see a fully pilotless passenger aircraft.

What I meant was that 95% of the way in a typical flight is fully automated, not that 95% of flights are fully automated. The pilot sets the flight plan to the autopilot and the plane flies itself under the watching eye of the pilots (hands off). They mainly need to make minor adjustments per air traffic controllers instructions (and those are not done by stick inputs but by altering the autopilot plan). The only two times the pilots actually fly the plane are the takeoff and the landing. About 5% of the typical duration of the flight.

Few people know that the autopilot is actually (albeit in a crude form) a very old invention. The first successful autopilot was trialed in 1914 by Elmer Ambrose Sperry.

Current autopilots are not designed directly for autonomous flight but it's been proven that even a current autopilot can actually land the plane successfully in good conditions.
 
We should really be saying, okay unmanned flights, IF you give us a cheaper flight...

If we're going to have unmanned flights then we should just get rid of the damned airlines in the first place and let me buy an unmanned drone big enough to carry me where I want to go. I'd rather have that than an unmanned car or sitting in a cramped tube filled with annoying smelly people with their crying kids.
 
Sauce or bs, I seriously doubt every jumbo jet pilot is a Maverik, I really do.

Not what I meant. I meant the engine out landings are more common than you would think. The reason you dont hear about them is because when they happen they are rather uneventful and not "sexy" enough for news coverage. Not I am talking about all of aviation here not just the airlines which most people think is "all" of aviation. Its so much bigger than youd realize.

No I'm not. Just a semi-enthusiast and simulator pilot.

Please go take some flight training and learn a bit. It will change your view and your entire world.

What I meant was that 95% of the way in a typical flight is fully automated, not that 95% of flights are fully automated. The pilot sets the flight plan to the autopilot and the plane flies itself under the watching eye of the pilots (hands off). They mainly need to make minor adjustments per air traffic controllers instructions (and those are not done by stick inputs but by altering the autopilot plan). The only two times the pilots actually fly the plane are the takeoff and the landing. About 5% of the typical duration of the flight.

Few people know that the autopilot is actually (albeit in a crude form) a very old invention. The first successful autopilot was trialed in 1914 by Elmer Ambrose Sperry.

Current autopilots are not designed directly for autonomous flight but it's been proven that even a current autopilot can actually land the plane successfully in good conditions.

Perhaps you should defer to those of us with qualifications on the subject. I have spent many hours studying aircraft and actually flying them. I teach this subject as well as earn money doing it. When I tell you that you are mis-characterizing what you have heard/learned I am not being mean I am trying to help you understand the reality of the situation as seen/experienced by those of us directly involved in the industry.

Yes some autopilots are capable of handling all phases of flight under certain circumstances. But most airliners are not sufficiently equipped and most airports are not either. Just because a pilot isnt holding onto the yoke doesn't mean he or she isnt flying the aircraft. Just as a previous poster pointed out a TCAS avoidance maneuver requires pilot input. That may take the form of directing the aircraft via a knob to turn left 10 degrees or grabbing the yoke and overriding the autopilot. Either way the human made the decision and applied control inputs - what those controls were is irrelevant to the discussion.

Thats not to say "stick and rudder" skills arent important - they very much are and the lack of them are contributing factors to some high profile accidents.
 
Computers just follow the rules as they are coded - every time, with no exceptions. It's one of the things that makes them such useful tools. But also only tools, and not in charge.

Computers are quickly gaining the ability to observe and analyze situations and make decisions based on previous experience and training. They're aren't just a ton of IF THEN ELSE statements and LUTs anymore.
 
Back
Top