http://news.yahoo.com/american-drone...cted-computer-virus-180019767.html

Just one of the many reasons (all involving safety) that pilotless - or even single-pilot - airliners are many, many decades away.

HAL

I have been very interested in pilotless airlines.

But, since my recent decision to become a pilot, this is probably the better outcome  

-CXfirst

This may seem peculiar, but I find it re assuring thast the pilot has a personal stake in me getting there in one piece.

I do not like the idea of pilotless airliners.

Quoting HAL (Thread starter): Just one of the many reasons (all involving safety) that pilotless - or even single-pilot - airliners are many, many decades away.

While I don't support pilotless airliners, what is stopping a virus from being in the flight computers of one of today's airliners?

Quoting poLOT (Reply 3): While I don't support pilotless airliners, what is stopping a virus from being in the flight computers of one of today's airliners?

If it happens, you've got two pilots to deal with the problem. The impetus behind single-pilot airliners is that the computers will be able to do the work of the pilot who is replaced. If those computers are compromised, then things could easily spiral out of control.

-Mir

Interesting that the OP's name is HAL...

Mir, what happens if the virus is in the FADEC and takes over control from he crew?

I think the discussion about pilotless aircraft is actually two discussions - remote-controlled vs autonomous.

Remote controlled: works when you assume you are the single occupier of airspace. At some point we will probably have remote controlled freighters.

Autonomous: Our computers are Turing-Complete machines, which means they are not capable of certain types of problem-solving that humans are much better at. The real rise of the machines happens when we are able to create machines that think like us. That journey hasn't even begun.

Quoting HAL (Thread starter): Just one of the many reasons (all involving safety) that pilotless - or even single-pilot - airliners are many, many decades away.

This was a keylogger on the *ground* computer, nothing to do with the drone itself. This is analogous to having a virus on the dispatcher's computer, rather than on the airplane. It's a complete red herring relative to aircraft onboard automation.

Quoting poLOT (Reply 3): While I don't support pilotless airliners, what is stopping a virus from being in the flight computers of one of today's airliners?

There's no fundamental technical barrier, it's just almost impossible from a practical standpoint. Even if you could get physical access, the amount of technical knowledge to actually write a successful virus for a custom OS running custom software with custom security...the only people who could do it are the people who made the flight computers themselves.

Tom.

Quoting HAL (Thread starter): Just one of the many reasons (all involving safety) that pilotless - or even single-pilot - airliners are many, many decades away.

Not sure what this has to with anything related to pilotless aircraft in the future??
So far, pilots have caused more fatal crashes that this 'virus'!

Quoting comorin (Reply 5): Remote controlled: works when you assume you are the single occupier of airspace.

ummm, why?

Quoting comorin (Reply 5): which means they are not capable of certain types of problem-solving that humans are much better at.

Which of these are humans better at, that a computer couldn't be ... and much faster?

Quoting comorin (Reply 5): The real rise of the machines happens when we are able to create machines that think like us.

...oh, and make the wrong decisions I guess? You are right! Computers will only be able to mimic the flying of humans when they make mistakes  


Jimbo

Quoting CXfirst (Reply 1): I have been very interested in pilotless airlines.

Me too.. I'll be down here on terra firma watching.   There's either 2 up front, or I'm walking.

Quoting tdscanuck (Reply 6): There's no fundamental technical barrier, it's just almost impossible from a practical standpoint. Even if you could get physical access, the amount of technical knowledge to actually write a successful virus for a custom OS running custom software with custom security...the only people who could do it are the people who made the flight computers themselves.

Yes, and I would expect that pilotless aircraft would have many of the same practical barriers.

Well with humans you get things like CFIT (controlled flight into terrain) - one of the most common causes for accidents.

Yes computers and technology might not be perfects, however I strongly suspect industry will be able to make automated flight as safe if not safer than with humans.

Human error sadly is an ever present and quite significant danger in aviation, and frankly if it reduced or replaced might be quite beneficial.

I'm just thinking in the age we live in with the terrorist and such this is not a bright idea. I don't think I could fly on such an aircraft. I know humans make errors, but humans can also think for themselves. Computers can only do what they are programed to do.

Quoting HAL (Thread starter): http://news.yahoo.com/american-drone...cted-computer-virus-180019767.html Just one of the many reasons (all involving safety) that pilotless - or even single-pilot - airliners are many, many decades away.

http://www.cbc.ca/news/world/story/2.../14/buffalo-crash-probe051409.html

During his testimony Thursday, Robert Key Dismukes told board members that flight data and transcripts of cockpit recordings show Capt. Marvin Renslow failed to realize the plane was about to stall as it approached Buffalo Niagara International Airport in wintry weather.

"I don't see any evidence he knew the situation he was in," said Dismukes, a pilot and scientist at NASA's Ames Research Center who has spent decades studying fields such as pilot attention and cockpit distractions.

One of the many reasons we should remove pilots.   

No I do not think we should. Just saying there is a lot more to it.

Quoting cmf (Reply 12): One of the many reasons we should remove pilots.

I'm not trying to say that automated aircraft are useless. Nor am I saying that human pilots are inherently better than automated ones. But there have been several threads here recently where a few people have been very vocal about going single-pilot now, and pilotless in the future, all in the name of 'removing the unsafe link' in the safety chain.

My point in posting this story is to show that there will always be problems with automated (and remote - in the case of pilotless aircraft) flight systems. What is best, and what will be safest, (which has been my point all along), is a mix of the two. Improve pilot training so that areas that are becoming lax in the advent of more automated aircraft, such as the stall recognition & avoidance problems from the Colgan and Air France disasters, and design better human/machine interfaces so that both sides of the instrument panel can monitor the other and make suggestions/corrections to keep the flight safe.

Does it cost more than an old-school analog cockpit with two pilots? Does it cost more than a purely automated aircraft? Is is safer than either one of those? Yes, on all counts.

HAL

BTW, for comorin, my screen name comes from the airline I work for, not for the legendary malfunctioning computer.

Hal,

We probably agree for the most part but to expand.

Automated systems will always be safer and more consistent WHEN we have the knowledge to provide them with information to handle all (virtually) situations they will experience. As it stands in aviation we have the knowledge to do that for the vast majority of situations but not enough to entrust them with human lives.

Problem is we are in a catch 22 situation. Automated systems make less errors under normal conditions but are essentially unable to handle exceptional situations. Humans are becoming less able to handle the exceptional situations because the are not getting the experience from flying under normal situations.

Probably we should let pilots train under normal conditions without passengers but that is economically unrealistic.

We need to pick a poison but no matter which is picked there will be people with good arguments for why it is wrong. Lose - lose  

Quoting HAL (Thread starter): Just one of the many reasons (all involving safety) that pilotless - or even single-pilot - airliners are many, many decades away.

As much as I agree with your posts HAL I find this thread quite irrelevant to the matter of pilot-less aircraft.

As has been pointed out, it was the ground computer that has the virus. From the sounds of it, I'm guessing it's just a program that's meant to be there that nobody realized was actually there before. Quite silly IMO. And at worse it's just a keylogger. There's no indication of any information being sent anywhere. This does not surprise me one bit, as the ground computers are running commercially available operating systems. Something airplanes do not have.

Quoting tdscanuck (Reply 6): It's a complete red herring relative to aircraft onboard automation.

Indeed. As always a fine piece of aviation journalism.   

I remember a thread in tech/ops about a similar subject. Basically the conclusion was is that its pretty much impossible to put a virus on say, the FMS system. These all run on embedded, real-time, extremely proprietary operating systems which only a couple of hundred people have access to the code to begin with. Not to mention it's not like you can just plug in a USB port and voila, cause an infection (though some newer FMSs may have a USB port)


EDIT: found the thread. It's very informative: Aircraft Virus Vulnerability (by Ferroviarius Sep 25 2010 in Tech Ops)

[Edited 2011-10-08 14:44:24]

Quoting HAL (Reply 13): BTW, for comorin, my screen name comes from the airline I work for, not for the legendary malfunctioning computer.

Mr. Clarke once said that he came up with the name HAL because he wanted the computer in 2001, A Space Odyssey to be one better than IBM. Hence, one letter to the left.

HAL, Im with you on this one. I happened to have read the article this morning. Even tho I have been around airliners all my life, and computers do wonderful things. The old adage applies : Garbage in, garbage out. We are a long, long way from pilot-less a/c. Even then, I believe there will always be a need for people to have some input.

Quoting cmf (Reply 14): Problem is we are in a catch 22 situation. Automated systems make less errors under normal conditions but are essentially unable to handle exceptional situations. Humans are becoming less able to handle the exceptional situations because the are not getting the experience from flying under normal situations.

I agree. Allow the computer to operate under normal situations. In the event of a situation developing, the pilots can take over.

JD CRPXE

There is a quote from the 'Dreamland' book series, I haven't been able to find it again word-for-word, but it essentially said that for aircraft, the accuracy and precision of computers will never be able to overshadow and/or replace human ingenuity and creativity. Might try to come close, but "try" will ultimately be the restraint on any such attempts until we achieve true AI.

HAL also was a short form for Heuristically Programmed Algorithmic Computer. Hopefully HAL the poster is not a computer masquerading as a person!  

My more serious and nerdy point was trying to point out the limitations of today's automata. It has bearing on both UAVs and Automation in the cockpit. Computers today are theoretically only capable of doing what is programmed into them - whereas the dumbest chimp is capable of solving spatial problems. Sure, you have fuzzy logic, pattern recognition, neural networks and inference, but AI has stopped there for very good reason. Computer scientists are realizing the fallacy of equating human intelligence with computing power, when they are actually different things.

So computers make great machines - autopilots, FADECs and so on, and their reflexes are better than any pilots. They are great at taking over workloads - same since Archimedes time. But computers are only as good as their instructions, and humans bring in skills that are not replicable in today's machines. Computers have also become very good mimics of humans, as anyone who has played a computer game or been in a simulator knows.

We don't have a clue yet as to what constitutes sentience, intelligence, consciousness and the mind. When we find out, and we can build organisms that can think, then the future is truly upon us.

In passing, I leave those interested with a link for background on the limits of Turing-complete machines:

http://www.npr.org/blogs/13.7/2011/0...limits-of-turing-complete-machines

Quoting bond007 (Reply 7): Which of these are humans better at, that a computer couldn't be ... and much faster?

Actually this is the problem with pilotless or one-pilot A/C: Computers do not have judgement skills; they can only problem-solve based on a finite amount of data or variables. Flying can potentially present an infinate number of variables, so until computers start to think the way we do, the way they function cannot compare to the way the brain functions. No, humans are not infamible, but our brains are still incredibaly more complex than a computer's. I suspect the time will come, but I don't think we are anywhere near there yet.

Quoting HAL (Thread starter): Just one of the many reasons (all involving safety) that pilotless - or even single-pilot - airliners are many, many decades away.

I think computers are more reliable pilots than their human operators. I doubt many commercial airline flights are even hand flown anymore. Indeed the amount of automation you fear coming to fruition is already a reality. Human error nowadays seems to be the primary cause of fatal commercial airliner crashes. It was for AA 587, Colgan 3407 and AF 447. And that's just an excerpt. I have yet to hear of a properly programmed FMC flying a planeload of people into the side of a mountain or straight into the ocean, but I can give you a laundry list of pilots who have done just that.

I certainly think that this thread, raises valid concerns, but is laden with ignorance.

Quoting flybyguy (Reply 21): Human error nowadays seems to be the primary cause of fatal commercial airliner crashes. It was for AA 587, Colgan 3407 and AF 447. And that's just an excerpt. I have yet to hear of a properly programmed FMC flying a planeload of people into the side of a mountain or straight into the ocean, but I can give you a laundry list of pilots who have done just that.

Just food for thought, but this is the logical fallacy that so many supporters of completely automated aircraft believe. They say 'Human errors caused these crashes, so humans must be eliminated from the loop'. The logical problem you've created is that yes, there are a few human caused crashes. But how many other accidents, incidents, and disasters have humans avoided with a simple glance, thought, and flick of a switch or pull of a control? How many of those unknown accidents would not have been avoided if it were only a machine piloting the plane? And no, a properly programmed FMC has not flown a plane into a mountain or crashed into the ocean. But if you automate the plane, who is going to program the FMC? Some human has to enter the planned route, whether a human or machine is flying the plane! That person can make a mistake whether he is in the cockpit, or in a remote control facility. The difference is that if a mistake is entered, the pilot in the cockpit has a better chance of detecting and correcting the mistake than the guy separated from the aircraft by thousands of miles of distance and is connected to it only by radio waves. And if it's only a machine operating the plane, how is it going to be programmed to detect a mistake when it is only doing what it's told to do by the FMC? Yes, I'm being sarcastic with that last one, but you get my point.

Today we have an immensely safer aviation system than in the past. Look at the accident record from past decades, and you'll see that we're doing an amazing job today. That's why the few crashes that do occur, create such a frenzy - because they really are so rare. The easy causes have been discovered, and now we are left to root out the subtle and difficult-to-find reasons that were ignored for so long.

The argument that human pilots cause crashes and should be eliminated, ignores the fact that human pilots are extraordinarily good at detecting and avoiding situations that can lead to crashes in the first place. Look at the accident rate and causes from the last 50 years, and you'll see a lockstep increase in safety, not only from improved technology & systems, but from improved training based on real-world accident investigation, human factors studies, and advanced learning techniques. Both better trained pilots, and better technology are what's needed to improve safety. Eliminating one or the other would only lessen the high level of safety we enjoy today.

HAL

As long as "certain" countries develope destructive software like stuxnet, even planes with cockpitcrews on bord are unsafe because there is an permanent, potential threat. Newer planes are much more addicted of computertechnology - so there might be an additional, higher risk. But also other essential systems for aviation are in danger. To give some points: Fueling, Airport infrastructure illumination, ground navigation systems, communication systems etc. etc. etc. . Concerning all these vulnerable points, pilotless planes would be a piece of cake in such virus szenarions. However, two things are facts. 1st: pilotless airliners would be more safe than todays airliners - for the pilots... 2nd: Due to psychological reasons, pilotless airliners won't be accepted by an majority of passangers. So don't get nervous. Even in 50 years, airliners will have pilots on bord.

Quoting HAL (Reply 22): And if it's only a machine operating the plane, how is it going to be programmed to detect a mistake when it is only doing what it's told to do by the FMC?

Who said pilotless aircraft would simply be flown by the FMC? Yet another 'logical fallacy'! Pilotless aircraft would have completely different systems from current designs. Today's FMCs etc. are designed around the fact that there is a pilot or two, in the cockpit.

Quoting HAL (Reply 22): ignores the fact that human pilots are extraordinarily good at detecting and avoiding situations that can lead to crashes in the first place

So, let's have an example where a pilot made a decision that a computer could not have done. We've had this discussion a few times on this forum, and nobdoy has come up with a good example.

'Seat of the pants' flying simply does not exist, and those that have tried it sometimes have bad consequences.


Jimbo

Pilotless airliners would not really be pilotless. Just like the military drones, they would have pilots flying them by remote control from the ground. The airspace system is too dynamic to be able to simply program a flight and have it go off without any changes that have to be managed real time.

Quoting bond007 (Reply 24): So, let's have an example where a pilot made a decision that a computer could not have done. We've had this discussion a few times on this forum, and nobdoy has come up with a good example.

That's fairly easy... A pilot can look out the window and say "given my aircraft performance and payload, I don't like the condition of that runway and I will do X, Y, and Z to compensate". A computer cannot, at least today, judge runway conditions.

Quoting bond007 (Reply 24): So, let's have an example where a pilot made a decision that a computer could not have done. We've had this discussion a few times on this forum, and nobdoy has come up with a good example.

US1549. I doubt any computer could have made the decision of not attempting to return to an airport (LGA or TEB) but looking out the window for other options, evaluating them and selecting the Hudson as the best option.

Quoting ZANL188 (Reply 26): That's fairly easy... A pilot can look out the window and say "given my aircraft performance and payload, I don't like the condition of that runway and I will do X, Y, and Z to compensate". A computer cannot, at least today, judge runway conditions.

Of course it can, and they do all the time.

Quoting golfradio (Reply 27): US1549. I doubt any computer could have made the decision of not attempting to return to an airport (LGA or TEB) but looking out the window for other options, evaluating them and selecting the Hudson as the best option.

OK, but why not ... the decision was made based upon a number of factors - something computers are very good at doing,and doing very quickly and accurately.


Jimbo

Quoting bond007 (Reply 28): Of course it can, and they do all the time.

Please explain. I have yet to see a computer judge runway conditions, report them yes, make a judgement no.

Quoting bond007 (Reply 24): So, let's have an example where a pilot made a decision that a computer could not have done. We've had this discussion a few times on this forum, and nobdoy has come up with a good example.

The not so uncommon occurrence of AP giving up and telling pilots to take over because it can't make sense of the inputs it is receiving.

Quoting bond007 (Reply 28): Of course it can, and they do all the time.

Show me ONE case of a computer telling the pilot to land on a different runway than the one programmed, and the reason it made that decision, and I'll buy you lunch.

Quoting bond007 (Reply 28): Quoting golfradio (Reply 27):US1549. I doubt any computer could have made the decision of not attempting to return to an airport (LGA or TEB) but looking out the window for other options, evaluating them and selecting the Hudson as the best option. OK, but why not ... the decision was made based upon a number of factors - something computers are very good at doing,and doing very quickly and accurately.

Your profile says you are a pilot, so I'm having a hard time understanding how you can claim this. How in the world would you program a computer to make that kind of decision - to make the leap of reasoning that it's better to put it down in the river, rather than try for the nearest runway? What if there were an empty freeway next to the river? What if there were were a long grassy park next to the river? Would the programmers have had the foresight to program the computer to make those kind of decisions? What if the computer aimed for the park, but it was full of kids playing sunday soccer? Would the plane have the ability to see the difference between an empty park and a busy one, and know in seconds whether it was the best place to land? Of course not. But the human eye and brain can take those kind of inputs and make an immediate decision on what is best. There is no quesiton that both human and computer need to work together inside the aircraft, to make the best and safest flight possible. Anything else is either simply fantasy, or a reduction in safety.

HAL

Quoting golfradio (Reply 27): US1549. I doubt any computer could have made the decision of not attempting to return to an airport (LGA or TEB) but looking out the window for other options, evaluating them and selecting the Hudson as the best option.

US 1549 is one of the instances in aviation history that I seriously doubt a computerized pilot would not be able to execute. However, I do think there is more chance for human error in the many, many times when things are routine than in the statistical improbability that things go wrong. Personally I would take the risk of not having a pilot on board in the event of an emergency simply because there is no guarantee that a person would bring an airplane down successfully when a computer cannot. I really think we have the technology today to fly passengers in full automation, but public perception of combining flying with automation is still profoundly negative.

Yet, people won't mind driving a car in unserviceable condition on the highway or (from personal observation) getting into a car with a driver that's buzzed with alcohol.

US 1549 may not have even been successfully duplicated by other seasoned pilots either. That instance was just having the right circumstances falling into place. Had it been summertime, the Hudson would be filled with personal watercraft. Had it been IMC Sully wouldn't have been able to see a place to set the plane down. Had it not been for the experienced, senior crew, the situation would have turned out to be a tragedy.

Pilots nowadays train in the most advanced flight simulators money can buy. They train to resolve most foreseeable emergent issues through a series of procedural checks and gates to result in an expected outcome (ex. engine out on take-off). It's hard for me to believe that a computer couldn't run through these sorts of checklists in fractions of a second. Certainly a computerized emergency resolution protocol can be executed without human factors that may inhibit successful mitigation of danger to the aircraft better than a pilot can. Do I think it's perfect, no... but it's a start in the right direction.

My entire argument is to debunk the unreasonable expectation that air travel should be 100% safe. It wasn't yesterday, it isn't today, and it won't be tomorrow. We can all live in fear of possibilities or simply live our lives FOR possibilities.

I think technology has gotten us close to that 100% safety margin and perhaps removing human pilots from the flight deck will get us even closer to that number someday... but 100% safety in anything is just a pipe dream.

Quoting HAL (Reply 31): Show me ONE case of a computer telling the pilot to land on a different runway than the one programmed, and the reason it made that decision, and I'll buy you lunch.

Computers are used to calculate runway length needed and in that way may tell the pilot to land on a different runway. Reason is, of course, that entered parameters will place it outside acceptable parameters.

The reason for using computers to do the calculation is right up the core of the argument.

I've never thought very positive of the idea of a pilotless cockpit. It is just dumb to not have a thinking human being behind the controls.

Quoting bond007 (Reply 28): Of course it can, and they do all the time.

I've yet to see an FMS or any other computer on an aircraft that tells me the runway is contaminated. Thus far I've always been the one to let the airplane know about runway conditions.


A question for the pilotless fans, there have been several times I've seen an aircraft not on our TCAS that we would have run into had we not been up front to see the airplane and avoid it. There have been a handful of occasions we have elected to discontinue an approach, varying from turbulence to breaking out and not feeling comfortable with the runway conditions (unplowed even though it was reported as plowed, or we saw an airplane taking the runway or on a crossing runwauy on the roll...you've just got to love uncontrolled fields) and elected to go missed. During winter ops we often elect to return to get deiced again because we visually see fluid failure, or snow and/or ice accumulating on the aircraft again. I've shut engines down after we started them at the gate when wayward rampers or a tug with a bad parking brake walked or rolled towards our running prop. To the best of my knowledge these situations would not have been avoided without a pair of eyes and the judgement of an experienced crew up front. How do you think a computer would handle these situations?

My experiences are not entirely unique, while details vary, I have talked to several crews who have encountered situations in which human judgement was the difference between a safe outcome and a leading news story.

[Edited 2011-10-09 13:53:50]

First of all, and this happens in all pilotless aircraft discussions, folks think that pilotless aircraft means an A320 with a flight computer, flying into runways that look just like they do today, and in airspace just like today. If pilotless aircraft ever happen, then all of the above would need to change... We're not talking about just removing the pilots - we are talking about aircraft, airspace, runways and airports all designed specifically for pilotless airplance.

Quoting HAL (Reply 31): Would the plane have the ability to see the difference between an empty park and a busy one, and know in seconds whether it was the best place to land? Of course not. But the human eye and brain can take those kind of inputs and make an immediate decision on what is best

Of course they can... and make the decision faster. Cameras in space can identify and recognize the faces of folks ... the technology is already here for that. As mentioned by other posters, the Hudson river argument might be the only one that 'may' have a good argument, but those decisions were still based upon sensory inputs .... not some seat of the pants magic.

Quoting futureualpilot (Reply 35): How do you think a computer would handle these situations?

Hundred or thousands of sensors, not only in the aircraft but on the runways. Good grief, my car automatically tugs my seatbelt if a car cuts in front of me, and keeps a pre-set distance behind traffic and will actually brake the car to a stop .... so,not sure why you think any simple system couldn't tell if the runway was blocked, or icy, or whatever .... radar can pick up a bird ... it could easily pick up a Cessna 172, as could any camera in VMC ....

Remember what I said earlier ... we aren't talking about today's airplanes!


Jimbo

I'm not a pilot, so I don't have a ton of insight into that side of the equation. But I am a computer programmer, and I'm guessing most pilots don't know as much about theories of computing and programming as I do.

One point to be made is that most of the things that you are saying "computers can't do" is really "computers right now can't do this." A computer mainframe from the 1960s couldn't play youtube videos. Does that mean that youtube videos can't be played on computers? No.

Moreover, a lot of things that are said such as "a computer couldn't determine runway conditions" shows a terrible lack of knowledge of things such as AI and Image processing. Why couldn't you have cameras installed at airports that computers analyze the pictures from to determine the runway condition? I think if you did some research you would be surprised at some of the decision making AI is capable of today. Who knows what the state of AI is going to be in 40 years.

Now, you are absolutely right that in today's environment, with today's airplanes, and today's ATC, and today's assumptions about how flight works in all stages of flight that pilotless or 1 pilot flight is not going to work. What that doesn't mean is that it is never possible given revamps and technological changes in all areas.

I'm also not saying that it will happen, or ever should. But the argument that the FMC is your airliner right now isn't able to determine if a runway is landable, or able to make decisions about what to do in case of X is not the same as saying that the same thing is impossible.

Quoting bond007 (Reply 36): Hundred or thousands of sensors, not only in the aircraft but on the runways. Good grief, my car automatically tugs my seatbelt if a car cuts in front of me, and keeps a pre-set distance behind traffic and will actually brake the car to a stop .... so,not sure why you think any simple system couldn't tell if the runway was blocked, or icy, or whatever .... radar can pick up a bird ... it could easily pick up a Cessna 172, as could any camera in VMC .... Remember what I said earlier ... we aren't talking about today's airplanes!

After getting them approved by the FAA and figuring out how to install them, then the maintenance cost associated with ensuring they operate properly and accurately, I have a hard time seeing something like thousands of sensors at airports all across the country being financially viable. Not to mention making them tamper proof and setting guidelines for using them.

Quoting NathanH (Reply 37): Why couldn't you have cameras installed at airports that computers analyze the pictures from to determine the runway condition? I think if you did some research you would be surprised at some of the decision making AI is capable of today. Who knows what the state of AI is going to be in 40 years.

You could, but airports can hardly get approval for things like new runways and taxiway upgrades. Revamping the way our entire industry operates then actually implementing these things would make it almost certainly too costly to be viable and I doubt you'll get airlines to shoulder the bill for this sort of thing. Forget insurance premiums for a previously untested system being used on aircraft carrying passengers and getting general public acceptance for not having pilots up front.

Then you have the human aspect, tens of thousands of professional pilots whose very livelihood would be put in jeopardy and the unions they(we) belong to. This isn't like getting rid of the FE seat (which still exists even today) or going from props to jets. This is pilots being faced with the end of their profession.


I'm not saying it CAN'T happen, but I have serious doubts any of us will face pilotless airliners in our lifetimes, if anything, due to the potential costs associated with getting approval for, and implementing these systems.

[Edited 2011-10-09 14:24:38]

Quoting futureualpilot (Reply 38): You could, but airports can hardly get approval for things like new runways and taxiway upgrades. Revamping the way our entire industry operates then actually implementing these things would make it almost certainly too costly to be viable and I doubt you'll get airlines to shoulder the bill for this sort of thing. Forget insurance premiums for a previously untested system being used on aircraft carrying passengers and getting general public acceptance for not having pilots up front.

I agree with you. The difference is that thinking like that makes me incredibly sad. It makes me think that as a society we have decided to stop progressing. We are content with saying things like "We could do that but we won't because it is too hard or complicated."

And I don't want pilots to lose their jobs. I have a huge deal of respect for pilots and I love the aviation industry. But there are lots of groups that have lost their jobs for the betterment of society. It is something that happens as we progress technologically. So who knows what the world of aviation (or anything) will look like in 50 years? Maybe all of those pilots will be working from a bunker controlling 5 flights at a time. Or maybe they will still be up front of the planes.

My main point is that "can't now" shouldn't mean "can't ever".

Quoting bond007 (Reply 36): First of all, and this happens in all pilotless aircraft discussions, folks think that pilotless aircraft means an A320 with a flight computer, flying into runways that look just like they do today, and in airspace just like today. If pilotless aircraft ever happen, then all of the above would need to change... We're not talking about just removing the pilots - we are talking about aircraft, airspace, runways and airports all designed specifically for pilotless airplance.

Ah, I see you're in the realm of sci fi. Unlimited budgets too no doubt. Since you're dealing in sci fi why not skip airports and runways altogether? Flyvers, warp drive, transporters, etc. Endless possibilities in the sci fi universe.

Back in reality the rest of us are discussing using pilotless aircraft in the existing, or perhaps slightly modified, aviation system.

Quoting futureualpilot (Reply 38): I have a hard time seeing something like thousands of sensors at airports all across the country being financially viable. Not to mention making them tamper proof and setting guidelines for using them.

Being used to automating processes in manufacturing I do not think sensors will be much of an issue. Many of the critical sensors are already there, wind, temperature, many airports have sensors to measure friction and so on.

We would need to add sensors on the planes to identify activity around them but those exist in many other industries already and would actually be safer than hoping the pilots catch someone being too close.

Quoting NathanH (Reply 39): But there are lots of groups that have lost their jobs for the betterment of society.

A lot of professions have died out and been replaced with others. Didn't see many people who's work relied on Internet 25 years ago.

Quoting NathanH (Reply 39): Maybe all of those pilots will be working from a bunker controlling 5 flights at a time.

That actually scares me more than autonomous planes. The consequences of a link going down is something I do not like. The risks with someone deliberately blocking or worse taking over makes me sweat.

As long as we have subway drivers, I don't think we'll have UAV airliners.......

Quoting DeltaMD90 (Reply 42): As long as we have subway drivers, I don't think we'll have UAV airliners

There are places who do not have them. Decades already. People ride happily without concern.

Very different set of requirements to do it "human safe"

The number 1 reason we will never have pilotless airliners? Liability. No manufacturer or airline is going to want to bear all the responsibility for accidents. And with any accident, either the airline or the manufacturer would have an almost infinite amount of it. Every decision the computer makes will be within parameters specified by the airline or manufacturer. The aviation environment is just too dynamic to risk a company on that.

Quoting comorin (Reply 5): Mir, what happens if the virus is in the FADEC and takes over control from he crew?

Depends on the aircraft. In mine, you'd attempt a reset of the system, and if that doesn't work you'd have to figure out some other course of action.

Quoting NathanH (Reply 37): One point to be made is that most of the things that you are saying "computers can't do" is really "computers right now can't do this."

   I have no doubt that at some point in the future, we will see fully automated airliners. I do have doubts that I will see it in my lifetime (assuming the average human life span doesn't change too much). There's just too much stuff to work through in order to bring it up to standard.

-Mir

Quoting XJetflyer (Reply 11): I know humans make errors, but humans can also think for themselves. Computers can only do what they are programed to do.

But they can be programmed to do an awful lot more than most people realize. Make no mistake, there are certain tasks that humans are better at, and that we have no idea how to make a computer do. But those are generally not applicable to aviation. The *vast* majority of aviation problems are computable and, in most cases, better done by computers (hence the steady march of automation).

Quoting HAL (Reply 13): But there have been several threads here recently where a few people have been very vocal about going single-pilot now, and pilotless in the future, all in the name of 'removing the unsafe link' in the safety chain.

The claims that you do it to get safer are kind of hollow and the people you claim are very vocal about it usually aren't saying you should get rid of pilots to be safer, they're pointing out that removing pilots reduces *certain types* of accidents. It also introduces certain other ones. The requirement isn't that automation be 100% safe (that's a ridiculous argument), just that it can be equivalently safe.

Quoting HAL (Reply 13): My point in posting this story is to show that there will always be problems with automated (and remote - in the case of pilotless aircraft) flight systems.

Of course. And there will always be problems with piloted systems too.

Quoting HAL (Reply 13): Improve pilot training so that areas that are becoming lax in the advent of more automated aircraft, such as the stall recognition & avoidance problems from the Colgan and Air France disasters

I don't think you want to invoke Colgan or Air France in this context...those are textbook cases where a computer would have done far better.

Quoting HAL (Reply 13): Does it cost more than an old-school analog cockpit with two pilots? Does it cost more than a purely automated aircraft? Is is safer than either one of those? Yes, on all counts.

Yes to the first two...we have no certainty on the last one. We know that humans + automation is safer than humans alone. Nobody has ever done an airliner as automation only yet so claims about it's actual safety are just speculation.

Quoting cmf (Reply 14): Automated systems make less errors under normal conditions but are essentially unable to handle exceptional situations.

The're *extremely* capable of handling certain classes of exceptional situations, like control loss. Sioux City and the DHL freighter in Iraq were two marginally-survivable situations that would can be far better handled by increased automation.

Quoting ABQopsHP (Reply 17): We are a long, long way from pilot-less a/c.

We've got them right now. They're here. It's just a matter of when/how/if we decide to let them carry revenue passengers.

Quoting SSTsomeday (Reply 20): Computers do not have judgement skills; they can only problem-solve based on a finite amount of data or variables. Flying can potentially present an infinate number of variables, so until computers start to think the way we do, the way they function cannot compare to the way the brain functions.

Although it's true that computers don't think like we do, that difference is often their strength. There are obvious situations where they're weaker but, in the specific case of aviation where situation awareness is perhaps the single biggest safety factor, computers are much much better than humans because they can inherently multitask where humans cannot.

Quoting HAL (Reply 22): They say 'Human errors caused these crashes, so humans must be eliminated from the loop'. The logical problem you've created is that yes, there are a few human caused crashes. But how many other accidents, incidents, and disasters have humans avoided with a simple glance, thought, and flick of a switch or pull of a control?

But you're talking about systems design for joint human+automation. Nobody is suggesting you just slap a robot into a 777. You would fundamentally redesign an aircraft if it were designed for pilotless.

Quoting HAL (Reply 22): And if it's only a machine operating the plane, how is it going to be programmed to detect a mistake when it is only doing what it's told to do by the FMC? Yes, I'm being sarcastic with that last one, but you get my point.

I think you're missing the point...non-flight critical computers like the FMC are built to a far different standard than flight critical ones. You should be comparing to FBW, which has an appalling level of ability to detect, diagnose, and correct its own operation, to the point that a commercial FBW system has *never* suffered a complete failure in the history of aviation.

Quoting SRMD11 (Reply 23): As long as "certain" countries develope destructive software like stuxnet, even planes with cockpitcrews on bord are unsafe because there is an permanent, potential threat.

Stuxnet attacked completely unsecured, ubiquitous commodity-level COTS hardware. It's tough to imagine a hacker target more different than aviation.

Quoting golfradio (Reply 27): US1549. I doubt any computer could have made the decision of not attempting to return to an airport (LGA or TEB) but looking out the window for other options, evaluating them and selecting the Hudson as the best option.

The computer has access to a better terrain database than the pilot does and it works in all weather conditions. The issue of blockage is a thorny one, and is probably one of the biggest reasons you'll always want cameras and datalinks...but there's no reason that person needs to be actually on the aircraft. You can make a good argument that being a thousand miles away with a telephoto FLIR is a hell of a lot better position to be in than on the flight deck in the dark.

Quoting HAL (Reply 31): There is no quesiton that both human and computer need to work together inside the aircraft, to make the best and safest flight possible.

There's no question that human and computer need to work together...but why does the human have to be on the aircraft. The *only* input the human gets from outside that isn't filtered through the aircraft systems is visual and, thanks to NVG's, FLIR's, and lenses, you can actually get way better visual data if you process it through the aircraft anyway.

Quoting futureualpilot (Reply 35): A question for the pilotless fans, there have been several times I've seen an aircraft not on our TCAS that we would have run into had we not been up front to see the airplane and avoid it.

That's because current commercial aircraft don't have targeting radar...that's hardly a technical limitation. There are commercially available radar systems today that can track individual bullets; that's a hell of a lot better than any visual scan.

Quoting futureualpilot (Reply 35): I've shut engines down after we started them at the gate when wayward rampers or a tug with a bad parking brake walked or rolled towards our running prop. To the best of my knowledge these situations would not have been avoided without a pair of eyes and the judgement of an experienced crew up front. How do you think a computer would handle these situations?

Operational procedures around automated devices of any kind are always different than manned devices...this is nothing unique to aviation.

Quoting lowrider (Reply 44): The number 1 reason we will never have pilotless airliners? Liability. No manufacturer or airline is going to want to bear all the responsibility for accidents.

They already bear all the responsibility for accidents...whether its manned or not makes no difference.

Quoting Mir (Reply 45): Quoting comorin (Reply 5): Mir, what happens if the virus is in the FADEC and takes over control from he crew? Depends on the aircraft. In mine, you'd attempt a reset of the system, and if that doesn't work you'd have to figure out some other course of action.

The "FA" part of "FADEC" stands for "Full Authority." If the FADEC goes nuts there is, by definition, nothing you can do about it.

Tom.

Quoting NathanH (Reply 37): Why couldn't you have cameras installed at airports that computers analyze the pictures from to determine the runway condition?

Why use cameras? Have sensors (moisture or optical?) embedded in the runway, real time weather stations, and telemetry from the aircraft themselves. It isn't hard to envision a system where the weather station, aircraft, and runway itself are all uploading data in real time to a central airport computer to maintain a real time picture of the runway conditions and braking action.

Quoting tdscanuck (Reply 46): The *vast* majority of aviation problems are computable and, in most cases, better done by computers (hence the steady march of automation).

Even the best pilots are limited to only five senses, while a computer can utilize and keep track of a multitude of variables in real time.

Quoting tdscanuck (Reply 46): The're *extremely* capable of handling certain classes of exceptional situations, like control loss. Sioux City and the DHL freighter in Iraq were two marginally-survivable situations that would can be far better handled by increased automation.

   Some of the most interesting advances in controls (to me anyway) is the advent of planes that can re-learn how to fly after losing one or more control surfaces.

Quoting tdscanuck (Reply 46): The issue of blockage is a thorny one, and is probably one of the biggest reasons you'll always want cameras and datalinks...but there's no reason that person needs to be actually on the aircraft.

I'd argue that if such a system is now sufficient for a shoot/don't shoot decision, future systems will be more than sufficient for a land/don't land decision in what would be an exceptionally rare situation.

Quoting tdscanuck (Reply 46): There's no question that human and computer need to work together...but why does the human have to be on the aircraft.

They really don't. I think the first stage will be ground controllers overseeing (not doing on a regular basis, but overseeing) takeoffs and landings while the planes cruise almost completely autonomously, which they pretty much already do anyway. One "pilot" could fly dozens or hundreds of flights in a shift with a human still in the loop for the critical parts of the flight.

Quoting tdscanuck (Reply 46): They already bear all the responsibility for accidents...whether its manned or not makes no difference

No they don't, right now they can claim that their product was fine, it was the actions of the pilots or someone else that caused the accident or incident, thereby mitigating some or all of their liability. If all the possible actions are programmed at the factory as specified by the airline, then all accident causes will be reduced to a defective product or bad programming.

Quoting lowrider (Reply 48): Quoting tdscanuck (Reply 46): They already bear all the responsibility for accidents...whether its manned or not makes no difference No they don't, right now they can claim that their product was fine, it was the actions of the pilots or someone else that caused the accident or incident, thereby mitigating some or all of their liability.

But that never actually works. If it was flight crew error, the airline still bears liability (the flight crew were their agents), and the prosecution will always claim that faulty design lead to the error and get the OEM into the mix.

Quoting lowrider (Reply 48): If all the possible actions are programmed at the factory as specified by the airline, then all accident causes will be reduced to a defective product or bad programming.

And the airline will point and the OEM and the OEM will point at the individual programmer (it's still human error, just a different human), and the OEM and airline will still end up in court and settling.

Tom.