The stumbling block of this thesis is getting a computer to be smart enough to make good decisions in the complex, volatile, uncertain, and ambiguous environment that is travel by GA aircraft.

After all, flying a plane is easy - you can learn to take off, follow a GPS course, and land with only a few hours of instruction. The hard part is assessing risks in airplane, equipment, weather, and traffic, and managing those risks. Yes, the radar is showing a break in the line, but will it hold? Sure, it's a CAVU day, but the winds are forecast for 30+ gusty knots...is it okay to fly?

And that is the second stumbling block - will the owners of this autonomous airplane accept it when it refuses to fly because of ice en route? And what if something goes wonky (think the Eclipse auto-throttles a few years ago)? Do we maintain a cadre of ground-based pilots to take over when things go wrong?

What does that do to the economy and reliabilty of this system of autonomous airplanes?

Donald, if you haven't been watching the IBM/Jeopardy program airing this week, make a point to do so. I have to say, it is absolutely creeping me out.

It relates a little to this topic, in terms of artificial intelligence.

There is no doubt to me now that the technology will arrive very soon, whether or not we are ready to accept it is another matter. Probably getting rid of the FO in regional aircraft will be the first step. As for Donald's comments - the decision making aspect is exactly why the airlines and flight training operations are so much safer than GA business flying. When the emotion is gone and the go-no go decision is made for you (mostly), it is much harder to get in over your head.

Now does this stuff creep me out - you bet! So does my WAAS GPS at 250ft on a LPV approach - but I'd hate to be without it. I just hope they put a disconnect switch on these things so I can hand fly a little bit (but heck, that's what flight sims are for, right??)

I gotta ask... What auto pilot or ground-based human could have landed the Qantas 380 safely? Or do we decide ahead of time that anyone in an aircraft with a catastrophic failure (GA to airliner) are write-offs? Until the automation can get past major system sensor failures, there had better be an on-board intelligence that can use intuition and experience to bypass a lack of digital information.

I fly a Cessna 182 with G1000 and autopilot- about as automated as you get now in GA. I think with much of the work removed it actually makes flying more enjoyable. It also allows you to concentrate on often neglected parts of the process such as keeping a lookout for traffic.

The technology exists to replace AvWeb writers too.

We must assume that "auto-av" will be heavily GPS dependent ... kinda scary, considering the LightSquared proposals.

And shucks, if "kwazy kids" can wreak havoc on aircraft using hand-held lasers, wait 'till they get a taste of hacking into an aircraft's control frequencies.

Peter thats the guy in the corner of the flight ops room with the computers and they are there because they are good at flight sim. You asked :-)

In my former life I was in safety engineering in high hazard industries—chemical plants, refineries, nuclear power facilities, offshore oil platform, mines, etc. Beginning in the 80’s analog control systems started being replaced by digital control systems, going from literal steam gages (where do you think that term comes from?) to all digital all the time. These are some of the most complex and highly automated control systems on the planet. Aircraft systems are far, far behind in terms of system sensing capability and automatic fault handling.

As a result of automation, the number of operators (analogous to pilots) is now only a tiny fraction of what it once was but the sophistication and training requirements have gone up, way up. Why? Because even in the most sophisticated systems, operators must be prepared to assess, think, and act when things go wrong, go wrong, go wrong.

Continued. Pre-digital, controls system were simple and there were lots of operators only monitoring their little bit of the plant. Now operators not only have to understand the complex processes they are controlling, they have to understand the control systems controlling the process. They regularly participate in exercises like HAZOPS (hazard and operability studies http://en.wikipedia.org/wiki/Hazard_and_operability_study) that examine every component of the system—every valve, pipe, motor, sensor, pump, and ask: What are all the possible failure modes? What are the results of the failure? How will the failure be recognized? How will the failure be mitigated?

My point is this: heightened levels of automation increase the level of knowledge, skill, and training required; they do not decrease it. On any give flight, automated systems may do most or even all of the flying but the PIC still has to know what to do and when to do it when the environment is uncooperative or operating or control systems fail—as they inevitably will if one flies long enough.

How soon we forget! Think US Airways 1549 landing in the Hudson. Think Gimli glider. All those people would be dead if computers were in control of the flight.

Personalized fully automated flight is nearly here. I watched a video where Jay Carter of Carter Copters was told by someone from a major aviation museum that his Carter PAV will have such a huge impact on changing aviation that it'll be hanging in their museum between the Wright flyer and Spirit of St. Louis. Fully automated verticle take off and landing that I can keep in my garage, can't wait !

Meredith, those are good examples, esp 1549, where reaction time was so short -- if the airplane had to switch off to some remote emergency handler, it probably would have been too late. But by contrast, how many GA accidents could be avoided by automated systems, especially if there are safeguards built in to connect to emergency help when things go wrong?

P.S. I don't mean to imply that we are there yet, today, with the technology we have in hand. But I do think that from here, we can see these capabilities not far in the future.

Let me begin by saying I've enjoyed everyone's comments. I only wish that our Congress's minds were as great as those on this forum.

If your airplane can fly itself is it still fun to fly? Of course it is! Why just getting away from my wife is worth, no i'm just kidding honey.

One thing that is rarely mentioned about US Airways 1549 is the role of the flight systems. Airbus has far more automation than Boeing in their aircraft. I wonder what the outcome would have been without all that extra assistance. However, a big part of the thrill of flying is having your hands and feet on the controls. Nothing can replace that.

Wm Langweische wrote a book about the Airbus systems in relation to Flight 1549... you can read about it at this link, there is also a link thru to Paul Bertorelli's blog about it. http://www.avweb.com/avwebflash/news/SullenbergerFindsFaultWithLangweischeBook_201554-1.html

I fly because I love to fly and feel the airplane. Now with all the help and the pilot as another passanger, I would rather use a simulator and "fly" all kind of emergencies, with no risk and no geographic displacement, I fly 100% manual and love every moment.

A large part of dealing with mishaps on highly automated systems involves in-depth understanding of the very complex fault handling logic. For one example, nobody knows yet how much the fault handling itself may have contributed to the AF447 crash. I don't think anybody can say for sure if autonomous flight control will lead to more or fewer aviation accidents. What we should be able to predict, based on history and experience with other systems, is that we will certainly have DIFFERENT accidents.

When non-artificial intelligence,a well trained and experienced pilot, is too far removed from the aviation feedback loop bad things can and do happen. AF447 could very well turn out to be a technology induced accident, as well as the mid-air over Brazil between a 737 and Biz-Jet that were all flying just as the computers directed them too. Computers can fail in even more unpredictable ways than humans, In fact Neil Armstrong had to take manual control from a computer to make the first moon landing!

Thank you Mr. Kruger for your very insightful post. Most of us are probably old enough to be quite aware that no system is fail-proof, and that one day part of that system will fail. Yes we will have to be trained to deal with that failure, and will become not just operators (pilots), but system evaluators.

I think it is fine for unmanned military aircraft to go to war or support ground operations. This is because these operations are always life threatening and if there is no live pilot the loss potential is only financial. Personal transportation is a whole different kettle of fish.

I believe it will always be difficult to allow lives of passengers to depend on automation. Automatic systems to assist the human pilot in command are fine - so long as the human has the final word. It would also be nice if the human was able to fly without the automation - something that is doubtful even in today's airline environment.

Computer systems are getting more powerful all the time, but I don't think they are getting more reliable. Software is so complex it is difficult or impossible to prove it is correct. Even if you can build hardware worthy of life and death dependence the software will always be doubtful.

My decades in the computer business tell me software is getting generally less reliable rather than better. Software developers are multiplying like rabbits, but they are not getting better educations and they are not developing better professional judgment.

One more problem with the proposed "Auto" pilot is the notion that such a craft can request and receive clearances in the NAS. My understanding is that the proposed NextGen system includes dependence on 1940's simplex FM Voice communications between pilots and human controllers for clearances.

I can't imagine the FAA changing to an automated clearance communication system in my lifetime or while anybody else around today is still kicking.

I don't know about the rest, but if you look into NASA's SATS proposals, they are suggesting automated ATC. They were also advocating a T-style instrument approach with dual IAF's and dual missed approach procedures. Virtually all new RNAV approaches have the dual initial approach fixes and "T" configurations as suggested in the program, and I'm seeing more dual missed approach points. I'd guess we'll see automated ATC within 20 years.

Josh,

NASA and FAA are at opposite ends of the political spectrum. NASA is about propaganda that makes the USA seem advanced and high tech. FAA is a very conservative outfit dedicated to keeping aviation safe and their bureaucracy intact.

Just because NASA predicts something doesn't mean the FAA will implement it. Indeed it might be just the opposite.

Well Im among the group of GA pilots who have 2 axis auto pilots, and use them on most long flights, but I often hand fly just to stay current and for the fun of it. Lucky me, I fly an experimental, so the price of my autopilot is affordable as compared to the prices paid for "cert" units costing 5 times as much, and may not be as capable.

No, it won't be as much fun to fly. It's the challenge and feel of hand flying that does it for me. In IMC there may be a use for it.

But as for "easy pushbutton" flying, I know we need to widen the pilot base, but do you really want to be sharing the sky with people picked randomly from the Walmart parking lot? It's scary enough trying to drive next to them!

Well Im among the group of GA pilots who have 2 axis auto pilots, and use them on most long flights, but I often hand fly just to stay current and for the fun of it. Lucky me, I fly an experimental, so the price of my autopilot is affordable as compared to the prices paid for "cert" units costing 5 times as much, and may not be as capable.

Paul - I'm just pointing out on the approaches, and GPS, and NextGen they did.

The real concerns with highly automated aircraft are: How will an autopilot dependent pilot who has no recent practice hand flying the aircraft in instruments, respond to automation failure, or the necessity to hand fly the aircraft in instrument conditions?

When training policies of manufacturers, regulatory authorities and operators concentrate their training on how to operate the aircraft with the automation, and then encourage (demand) that the aircraft be flown with the automation for almost all operations, it is easy to see how pilots become autopilot dependent and autopilot complacent. Absolute reliance on automation leads to the complacency and the lack of aircraft monitoring that is associated with the errant NW 188 flight in October of 2009 as well as other accidents in 2009 including Colgan 3407 (fatal to 50), Turkish 1951 (fatal to 9) in Amsterdam and Empire 8284 where a simple flap asymmetry and a lack of airspeed control led to the destruction of the aircraft.

Continued: In each of the approach phase of flight accidents, basic airspeed monitoring and the lack of an application of basic flying skills all resulted in accidents.

The increase in automation in aircraft over the past several decades seems to have clouded a basic concept of aviation safety. That is: “Fly the Aircraft First”. With autopilot dependent pilots, the mantra is “let the autopilot fly the aircraft, no matter what”.

The industry now trains the pilots of automated aircraft are trained to "fly with the automation", not “Fly the Aircraft First”. There have been accidents since the beginning of the “automation era” associated with the unwillingness of crews of automated aircraft to take over hand fly the aircraft in situations where that is probably the safest course of action. A mind set that makes operation of the autopilot more important than the pilot being able to hand fly the airplane effectively is a safety issue that must be a confronted for the future of aviation.

I agree 100%, with Thomas Olsen I was flying at 26 k ft and my oxigen system failed passed out and because I was flying the aircraft the flight changed from straight and level to a diving, I recovered concience at 10k and straighted the airplane, If I were using the autopilot I will be history now.

Flying with autopilot and following the GPS magenta line, and you watching the nice airplane ride, why fly?

Thomas,

The concept being floated here is even worse than your scenario of rusty pilots letting the automation control the flight. It is to have no pilot at all and have the lives of the passengers depend on the automation never failing. This is the pipe dream NASA proposes in their PAV system.

I personally prefer to have no automation in my plane. However, I think there are times when automation to assist the pilot is fine. The notion that you don't need to learn to be a pilot to fly and you can put your life in the hands of the automation with no backup just seems like a nightmare to me.

The idea that a backup qualified pilot can be contacted on the ground to recover from an automation failure just extends the point. It requires the automation to be working in order to contact the backup pilot and follow his instructions. In this scenario a single electrical or software failure can only end in disaster.

To like this proposal you have to believe the automation is fool proof or the passengers are dispensable. The population is too high anyway . . .

At some point the automation software and cheap redundant hardware will make autonomous flight an everyday affair, presuming NEXGEN is delivered and works to its promised capability. If prior government history is any indicator, the aircraft systems will be ready way before NEXGEN is. Autonomous control and multi-vehicle deconfliction technology is racing ahead of behemoth government programs like NEXGEN and the technology may well render them moot at some point. The more likely scenario for future aircraft is we are all flying computer nodes that are internetworked and talking to each other and negotiating safe spacing. Airports would just be another stationary mode to negotiate with, and ATC as we know it today will no longer exist. Who is going to pay for this? On the airline side, reducing the cockpit to a single safety pilot will more than pay for the additional electronics. Business Jets will likely experience similar cost savings, while GA will most likely require some expensive refits for aircraft not designed for or delivered with this technology. One other point, even with the best, most reliable automation going, accidents are going to happen and you can war game 0.99999999 reliability all day long. The true test should be is the accident rate lower than it would have been without the technology?

John,

Your comment seems correct from a technical and mathematical point of view, but it doesn't deal with the psychological realm at all. "Fear of Flying" is all about lack of control. People must give up control of their environment and indeed their future to board an airplane as a passenger.

Giving control of your future to a pair of well qualified humans is one thing. Giving it to a pile of transistors and a mountain of software is an entirely different thing.

As a light plane pilot, I am confident I can keep myself alive with appropriate decisions and actions. My life and the lives of my passengers depend on me being correct. I can't imagine ever being confident enough in automation to ride in a plane with no pilot. Even if the probability of failure is small it is too big to bet your life on.

The capability and reliability of systems in general, and the intelligence (at least in terms of fault recognition and mitigation) will continue to improve in aviation as it has in the hazardous industries I mentioned above; it is just one more aspect of the digital revolution. Regularly letting the black box do the flying, as some in this thread seem to assume, is not synonymous with rusty piloting skills. It’s true that less hand flying tends to erode skills. To a large extent, that’s what sims are for—to aid pilots who don’t hand fly their aircraft enough to keep their situational awareness high and skills sharp. Fortunately the digital revolution is happening in simulators as well as the cockpit. Full motion flight simulators for light aircraft (Redbird et al) are free falling in price even as their capabilities go up. The methods employed to maintain proficiency change as systems change—but the common sense requirement to maintain proficiency remains.

The capability and reliability of systems in general, and the intelligence (at least in terms of fault recognition and mitigation) will continue to improve in aviation as it has in the hazardous industries I mentioned above; it is just one more aspect of the digital revolution. Regularly letting the black box do the flying, as some in this thread seem to assume, is not synonymous with rusty piloting skills. It’s true that less hand flying tends to erode skills. To a large extent, that’s what sims are for—to aid pilots who don’t hand fly their aircraft enough to keep their situational awareness high and skills sharp. Fortunately the digital revolution is happening in simulators as well as the cockpit. Full motion flight simulators for light aircraft (Redbird et al) are free falling in price even as their capabilities go up. The methods employed to maintain proficiency change as systems change—but the common sense requirement to maintain proficiency remains.

It seems incredible to me that we are considering planes that fly themselves when nobody seems to consider automobiles that drive themselves. Only one in one thousand US residents can fly a plane but a large majority can drive a car. That suggests that learning to drive a car is easier than learning to fly a plane (of course we already knew that).

So why are we talking about self flying airplanes already? I think it is a result of NASA and their pipe dreaming. These people have a long history of screwing up nearly everything they touch, and yet we like to pretend their dreams are so great we all share their vision. Well, I don't!

Let's wait until we can drive around in fully automated cars before we consider self flying airplanes for personal transportation. At least if the auto-car automation dies most people could manage to stop the car on the side of the road. In an airplane with dead automation most normal people can only figure out how to hold their head while they are summarily killed.

Well, we *are* driving around in fully automated cars... at least the geeks at Google have been doing it on California roads, in an experimental fleet. I agree it would be great to start there, the accident rate is so high -- that's one of the main motivators for Google's research, it could potentially save up to a million lives a year around the world, not to mention all the injuries.

But, that technology doesn't necessarily translate directly to GA, where the methods and challenges will be different.

As for a self-flying automated GA airplane, don't forget there are also parachute systems as a last resort.

Interesting discussion. Aircraft automation is still not as advanced as automation in automobiles. For example, the vehicle stability control system in my car provides much more control on slippery surfaces than I can and keeps me out trouble. The autopilot in an aircraft can't keep one out of trouble. Even a task as simple as reducing the throttle when an aircraft encounters turbulence is very difficult to program. Maybe this will happen someday but there will have to be a lot more programing to deal with all the variables.

Yeah, it would probably still be fun, but in a different way than manual flying. Whether or not that matters much depends on how important the strictly transportation aspect of the mission is.

Is it ever gonna happen? Maybe, but not on a large scale, or with the stellar results that some might envision. As Glenn K. said above, the accidents will be different; but as sure as we're all imperfect (including automation designers), they will happen, only with the system making the "stupid" mistakes, instead of the live pilot.

Regardless of accident prevention or causation, the great demand, mass production, reduced cost and so forth part won't materialize, imo. That would necessitate use by a large group of people other than those who already fly. This is like the driveable airplane, or flyable car for the masses idea in that respect, which doesn't work in part because air and land travel are two different worlds.

Airplanes are much more unforgiving than cars in the areas of weight and balance, weather, and mechanical failure to name a few. Automation either isn't going to be able to handle all of that (and more) effectively, or it's going to do so in a cost-prohibitive fashion, relegating the privelege of personal flight once again, to a very small group.

To answer the original question. NO. Having an autopilot, tcas, automatic autonomous radio comunicator fine. Getting in a flying box with a START button and a vanity mirror, BORING. If you can't fly upside down your not flying.

This is a great discussion. It just occurred to me, re-reading it, that perhaps the most telling evaluation of the weaknesses of a human/automation interface is not the content of these posts themselves, but the number of double posts on the thread. I'm just sayin'.

Au contraire, Meredith Hutto (“How soon we forget!). If computers had been in charge, Flight 1549 would never have been: it would have “seen” the birds ten miles away, recognized the threat, and altered course. Simple as that. If, however, it had happened, the computer would have done everything Sullenberger didn't: it would have instantly recognized that there wasn't a snowball's chance for a relight and immediately enacted its constantly evolving Plan B, in this case returning to LGA. It would have used the correct frequency (121.5 along with squawking 7700), declared a Mayday, advised who it was, what it was, where it was, how high it was, number of souls on board, its intentions, and approximate ETA It would have simultaneous advised cabin attendants of the situation and intentions and ETA. It would have landed – safely – back at LGA – as did every pilot that simed the experience. Same with the Gimli glider. The computer would never have permitted a TO. That incident happened because of repeated human error: the fuel guy didn't know metric from pounds, the pilots weren't paying attention, and someone didn't believe their gauges and pulled a breaker. Had computers been doing the flying, virtually none of the major air tragedies, including 9-11, would have taken place. All Sullenberger and Gimli did was prove a definitive case for removing the human factor, pilots especially, from commercial aviation.

Let me guess Dave, your an engineer? You paint a pretty cool picture of an autonomous aircraft but I think your leaving out the "human factor" when it comes to designing, constructing, maintaining etc. What would your robot do in the case of the Sioux City crash? Anyway I think the real question is would you enjoy flying your airplane if you were only along for the ride? The automation safety thing is just a pointless back and forth because no matter what you do planes will crash, and despite that, air travel on a major carrier is so safe its not worth thinking about. The commuter crash isn't an argument for more or less automation its an argument for a more rigorous widebody approach to safety. I would like to hear from more pilots about whether or not it would be fun to be a passenger, by yourself, in your robot plane?

I know that automation is coming and I have a few thoughts on it: 1. Aircraft will become like living organisms in that there will be pervasive affective and effective systems - sensing and doing. Air pressure sensors all though the wings, control surfaces etc. Sensors to monitor not only voltages and fluid pressures but also the integrity of wires, pipes, structural elements etc. And forget mere swing wings and trim tanks - the shape of aircraft will no doubt be more adjustable. In these ways, the electronics ability to sense and optimise will surpass humans'. 2. The ability for a machine to reliably outperform a well prepared human in those rare contingencies is a long way off. What I've learned thus far in life though is that every major advance always seems to include the loss of some capability or feature. I think news of the demise of commercial or military pilots might be premature but at some point, in the face of superb new technology, the rare occasion a human would perform better will be seen as almost ridiculous to allow for.

John,

I suppose Airplanes without human pilots are similar to fighter planes without guns. It was tried in the F4 Phantom and they found they had to add an external gun pod.

I think it is fine for unmanned military and observation planes to exist and do as much as they can. Still I think the day of the automated pilot carrying normal passengers is a really long way off. We all will go to our reward long before such a concept is accepted by normal people.

Let's not forget the political aspect of this venture. It may well be (coming soon to a country near you, as in USA) that planes that fly themselves will be the only way the TSA allows anyone to fly. Further, the TSA will perform strip/cavity searches on all occupants to make darned sure there is nobody on the flight who would have a China-man's chance of altering the proposed path for any reason whatsoever, no matter what happens. So, while a medical might no longer be required (good riddance), I still would not return to the air.

@ Robert: “What would your robot do in the case of the Sioux City crash?”. Without taking anything asway from Al Haynes & Co., assuming the bot were programmed not only to utilize differential thrust, it would also have been cognizant of ground effect and no flaps, it being the latter that were the critical elements that turned the landing into a careening fireball, it being a possibility that the result would have been different. “air travel on a major carrier is so safe its not worth thinking about”...? I must be the exception, because I sure do. And according to surveys, since being hijacked is the number one 'flier fear', so do a lot of others. But if the aircraft were under autonomous control, no chance of turning it into a cruise missile. As for being 'so safe', since 80% of accidents are the result of human error, the robot would reduce accidents by a further 80%. My vote goes to the bot. As for ensuring that the robot is bugproof, give kids a chance to 'beat the bot' with a $10 prize every time they crash the plane.

@ Robert: “What would your robot do in the case of the Sioux City crash?”. Without taking anything asway from Al Haynes & Co., assuming the bot were programmed not only to utilize differential thrust, it would also have been cognizant of ground effect and no flaps, it being the latter that were the critical elements that turned the landing into a careening fireball, it being a possibility that the result would have been different. “air travel on a major carrier is so safe its not worth thinking about”...? I must be the exception, because I sure do. And according to surveys, since being hijacked is the number one 'flier fear', so do a lot of others. But if the aircraft were under autonomous control, no chance of turning it into a cruise missile. As for being 'so safe', since 80% of accidents are the result of human error, the robot would reduce accidents by a further 80%. My vote goes to the bot. As for ensuring that the robot is bugproof, give kids a chance to 'beat the bot' with a $10 prize every time they crash the plane.

Dave, the Sioux City crash was caused by the failure of a triple redundant system through an unforeseen failure mode. The solution applied by the pilots was discovered on the spot. I don't think you can assume that solutions to un anticipated problems will be programmed into your robot. I am not sure what you are the exception to since I did not say that no one worries about air travel safety which you "sure do." My point isn't really about whether or not a robot would be safer, my point is that today flying on major carrier the risk is vanishingly small. I don't see what surveys that demonstrate how poor people are at assessing risk adds to the conversation. I don't believe you can use the phrase "no chance" when talking about a hypothetical technology especially when worms and viruses have been used to incapacitate or destroy computer controlled equipment. The real question is would you enjoy flying a plane such as the one you describe and if so what would be fun about it?

Robert: Respectfully, the Sioux City crash was not 'unforeseen' nor was the solution 'discovered on the spot'. Precedent with the same aircraft and the same situation and solution ran back nearly a decade. It's just that arrogant humans chose to ignore it. I agree that we cannot use the phrase "no chance", but we can, as they say, live and learn from the mistakes of others, and getting back to Sullenberger and arrogance, we've learned nothing from the myriad of mistakes made that memorable day. As for the 'real question', as you put it, no pilot ever made, at one or more point in his or her career, didn't wish they were on the ground. I would enjoy flying my plane a lot more if I knew that God was my co-pilot. Alternatively, something almost as smart, perhaps along the lines of the new iPad ap that can actually land a light plane.

Paul, I bet there will be a move to full-sized unmanned systems followed by some atrocious accidents/losses and then a move back to keeping people in the airframe - similar to the process with the F4s I think? I am reminded of the drones that have crashed because the operators could not hear that the engine had stopped. Maybe virtual reality systems will improve to the point that operators will sit in a control room but feel like they are in the cockpit? Either way it's not what I want to see.

Aside from the aviation aspects of this, being able to exert force without risking lives will change the decision matrix for future governments in potentially negative ways.

Apparently my last comment was deemed abusive or insulting so I apologize if anyone was offended, that was not my intent. I was attempting to satirize the the calumny that attends the age old man vs machine debate especially when engaged in hypotheticals. It seems like the least interesting of arguments yet this format somehow seems to engender the urge to respond. Very few people have responded directly to the original blog post's title probably because it is not nearly as entertaining as engaging in the argument over whether or not robot planes would be safer. I think this is a bit spurious but I suppose enjoyable in a sort of masochistic way. Since this seems to be the true substance of the thread then I would just like to say that I feel a robot can fly a plane better than any human, that if I had an emergency "fly me out of peril button" I would be happy. The paradox is that if I had control over the button the human factor would still be present and would to some extent invalidate the projected safety benefits of the robot.

I believe that while an automated plane might be safer with regard to the elimination of pilot mistakes it is still vulnerable to human error in the maintenance and design regimes, and may not in fact provide any real measurable benefit in terms of increased safety over the record of major carrier deaths per mile/deaths per flight achieved in the last 10 years. Further I think it is valid to be concerned about what new failure modes will be created by the new technology that will not be foreseeable. Of course I could be wrong because we are talking about the future and nonexistent robots! Dave you'll have to let me know what the precedents were for the sioux city crash, do you mean door failures leading to hydraulic failure? If I take what you say at face value let me ask, why would the designers of the robot be any less susceptible to arrogance and ignoring than whoever your talking about with the DC-10. You'll have to post a link to the Ipad app that can land a plane because to my ear that sounds , respectfully, vaguely trollish which pretty much accounts for my last post! (Along with the implication of the computer being almost as smart as god!) Of course it is distressing to lose fields where human capacity was preeminent, the loss of craft to automated machine tools for instance, but we should be realistic about human limitations.

Our familiarity with the "oops" parts of safety technology that is itself sometimes part of the accident chain should make us sceptical about idealized technology solutions; on the other hand it seems obvious that computers, (for lack of a better word), can easily do things that humans must acquire great skill to do, and that perhaps human problem solving in the face of unexpected events is less important then complete mastery of the expected.

John,

You present three very interesting ideas.

I hope you are wrong about starting passenger carrying with automation pilots before they are ready. Alas, your scenario is believable.

I'm not sure about your virtual reality and ground based operator for aircraft carrying live cargo. The problem is the time delay from the operator input to the aircraft response and information return. The further the operator is from the plane the more significant this is. The speed of light is an absolute limit here. This means the aircraft really must be running on automation with remote suggestions rather than absolute control from afar.

Your last idea is really scary. The political and military impact of unmanned killer planes is something we have today. We don't know how the government deals with this in the short term, and how our enemies will deal with it in the longer term. What scares me even more is the apparent fact that it is the CIA operating these killer warplanes rather than the military. They have more interest in secrecy than any other aspect of their activities including honor. Putting the lethal pilot in a remote location is a small change for warfare, but removing the military entirely and replacing them with spooks is a really significant change.

Robert,

I don't share your apparent automation "Worship". I don't think computers are better than humans at all things - just a few like doing binary arithmetic and making pre-programmed decisions.

Indeed, humans are much better than computers at some things. The first one that comes to mind is pattern recognition, and this is easily extended to threat recognition. Computers don't experience fear - an emotion that keeps animals alive in the wild.

One example of pattern recognition is the mother who can pick out her child on a crowded playground from 100 feet on a foggy rainy day with unerring accuracy. A computer can't even do this with a clear image, but the human can work with a foggy and partial image to recreate the whole in her mind.

I really believe it will be a very long time before an automation pilot will be competent enough to carry really precious cargo like ordinary passengers. I'm sure it will never get to that point in my lifetime because I wouldn't take such an outrageous chance to ride in such a craft even if others promised me it was as save as commercial airline travel.

Robert: as requested, here's that link to the iPad Auto-Landing App http://www.popularmechanics.com/technology/gadgets/reviews/the-auto-pilot-ipad-app. You Tube has a video of him demo-ing it at Oshkosh 2010 http://www.youtube.com/watch?v=wo7Lj29eo9M

Paul, I am sorry if you got the idea that I worship automation, thats simply not true. I tried to express my belief that computers are better at some things but not others than humans. I do not believe that an automated airplane would be safer statistically than a human operated one, I believe I said that explicitly. However I do believe that a computer can fly an airplane with greater precision than any human can. It can not react to problems for which it has not be designed and that is its critical weakness. The last sentence of my previous post is really an open question that I don't know the answer to. Im not going to get on any robot plane myself. Robot spacecraft, yes.

Dave, thanks for the link. I'm a big fan of X-plane, (Austin Meyer's Ouvre,) and it will be interesting to hear how it works in his new plane. When you originally posted about an Ipad that could land a plane it sounded so ridiculous because I imagined an I pad somehow landing a 172 with steam gauges. Magic! I see now most of the weight is carried by avionic stuff. I was alway wondering why the military didn't have a system with all their fly by wire hardware that could detect G-Loc and level off or climb to save an unconscious pilot until he could recover.

As a lifelong student and enthusiast of automobiles, I am aware that if we were as afraid of technological change as some highly traditionalistic pilots, we would still be driving cars where you have to manually adjust spark timing, and manually adjust fuel-air mixture.

In the days of the DC-7 and Lockheed Super G Constellation, you needed a flight engineer as well as the usual captain and first officer. And for overseas flights, you needed a navigator. During World War II, bombers like the B-17, B-24, and B-29 had a commander (pilot), co-pilot, flight engineer, navigator, and radio operator. Of course airliners don't need a bombardier, but bombers had them too. The fact that you had all these human crew members didn't diminish the captain's authority to make the final decision. Likewise, having the equivalent of a navigator (your GPS system), flight engineer (your automatic engine control system), and co-pilot (your autopilot) doesn't diminish YOUR authority or YOUR responsibility as pilot-in-command.

Before I sign off: in the world of ships, it has long been customary for a harbor pilot (a man of the sea who has intensive knowledge of the waters in and around a particular harbor) to navigate a ship in these hazardous areas. But the guy who was appointed as captain by the Navy or the shipping company, is still THE CAPTAIN. That's the way it should be, no matter how much of an "expert" your electronic systems become.