Boeing studies pilotless airplane

[confusedalot]

You missed the point. Nowhere is there any indication of controllerless ATC. You need some form of ATC with pilotless airplanes. So how are pilotless aircraft to be managed in complex airspace environments. No need for verbal communications anymore, but some form of super advanced CPDLC would be needed, it seems to me. Air transportation operates as a whole unit, so day to day airspace management also needs to be addressed in order for these machines to be viable.

[Rockie]

You guys are arguing that technology can safely operate an aircraft autonomously which in the environment we operate in would end in disaster very quickly. It’s far too dynamic. But that’s just part of it.

It is a dynamic environment. But on what basis are you concluding that it is TOO dynamic? Out of the many environments that autonomous vehicles are being developed to operate in, IFR flight is one of the most controlled.

What none of you consider is that physically flying the aircraft is just one small part of the job, and not even the part I really get paid for. I am paid for my judgement. I am paid to say “no” when I deem it necessary, and as CWE (who occasionally knows what he’s talking about) is fond of saying the hard part about flying is knowing when to do that. It’s hard to do in large part because virtually everybody else including your employer wants you to say “yes”.

Actually, everybody has considered that. And it's a point I agree with, if you'd actually read and take in previous posts on the subject. The whole point of additional automation is to move the human up the command level, from control manipulator, to indirect control, to "tactical" level planning, to strategic and operational level. Your statement is a perfect example of that. The decision about whether to launch a flight based on the state of the aircraft, ground systems, weather, etc. is a tactical level decision, as is the decision whether to continue the flight if conditions change or whether to divert due to a system failure, etc.

None of those decisions require you to be sitting, watching the localizer needle, ready to take control if it deviates. The reason you have to do THAT in your aircraft is because the systems in your aircraft are not reliable enough to be able to do the job safely without monitoring. They were never designed to be. The entirety of the aircraft was designed with the concept of the human pilot being in control and being the backup for many of the systems. But the creation of better fault monitoring and fault tolerant systems with their own sets up of back up systems gradually diminishes the need for that human monitoring, leaving the human more available for the tactical and strategic level decisions.

Do you fly IFR to busy places Posthumane? Have you ever experienced the mix of IFR and VFR in places like that such as instructions to “fly the Hudson”? Have you ever actually flown an automated aircraft and progressively scaled it back all the way to full manual flight because it was required? I’m guessing not.

You’re also still focused on manipulating the controls. You’re talking about moving humans up the command chain from one level to another, and failing to realize the pilot already is active and required at all those levels. You mistakingly think you can make strategic decisions that will eliminate tactical ones...you can’t. Saying “no” could be done long before the flight is scheduled to leave, or it could be made in the span of one second when you decide to abandon a landing because of what’s happening, or more importantly what you think will happen. That is just one of countless and unpredictable situations that requires judgement, which requires an active, trained brain.

Has real AI capable of that been invented yet?

One other thing you utterly fail to consider is the requirement for humans to deal with and decide on human issues. We fly humans around Posthumane, not computers. We frequently deal with human issues and conduct our flights accordingly.

Has real AI capable of that been invented yet?

[trey kule]

No, I do not think real AI has been developed yet and we could be decades from that time.

But real AI is not necessary to reduce flight crew by one. Or, possibly eliminate on board crew entirely. There is a huge dufference, for example, between cars and aircraft in many critical areas, that will enable aviation to adapt more quickly to technology

The signs are there if one looks. The push to allow single crew in cargo aircraft,,,A small step, but significant. Boeing looking into pilotless aircraft. Millions of dollars being spent because, it seems, they can not see what some can see ( cough,*** cough***, Rockie). with regard to human superiority.

It kind of harks back to a century or so ago, when experts pontificated that man would never fly , while the first planes were quietly being built in shops all over the world.

[Rockie]

"Boeing studies pilotless airplanes"

Flying cars and supersonic executive jets have been studied for decades. Studying doesn't in any way mean it's going to happen, it's just being looked at as are countless other things that don't materialize. Waiting for AI which, yes Trey Kule, is necessary before serious study of pilotless passenger aircraft can even begin.

The Boeing guy in charge himself said this:

"But a self-flying plane would need to be able land safely as Captain Chesley Sullenberger did in the "Miracle on the Hudson," Sinnett said. "If it can't, then we can't go there."

Substitute "landing on the Hudson" for pretty much anything else requiring critical thought, judgement, compassion, intuition, perception, prediction etc. The list of human abilities required that computers don't possess is really really large. Quoting the Star Wars movies, "I have a bad feeling about this" has saved the day in real life innumerable times.

[rookiepilot]

I'm beginning to think you don't fly airplanes and are therefore unqualified to comment on what it takes to be safe and effective flying them. You're an engineer right?

Unless I'm wrong, I suspect your background is flying aircraft, and not top level engineering.

I suspect the men and women working on autonomous driving, and autonomous aircraft one day, are in the same league as those engineers you like to denigrate, who designed the space shuttle, put men on the moon, and designed turbine engines with almost a microscopic failure rate, that you trust with your life.

In other words, geniuses.

Regardless of your flying qualifications, I'd be a little hesitant to speak with authority on what this group can and can't design to a high standard, given enough time and practical application.

I doubt they will ask you, frankly, or me, what can or cannot be done.

[Rockie]

Engineers work best when they’re supervised by subject matter experts telling them what their product is supposed to do. So if you don’t tell me what it takes to fly airplanes I won’t tell you what it takes to design high level systems.

[confusedalot]

Just saw a Nova documentary on PBS just last night about ''can we build a brain''. I'll spare you the details, but the answer is somewhere between ''no'' and numerous decades. It appears that even the most simplest of human activity that we take for granted becomes exceedingly complicated for current, even top of the line, computers. Well put together documentary.

Of particular interest I found were the self driving cars. Stating the obvious, they operate in two dimensions, lateral and time. An airplane operates in four dimensions, the basic three plus time. Like most people, thought that self driving cars could navigate any city any time, and avoid any mishap. It appears that they can only negotiate specific preplanned routes. They don't tell you that in mainstream media. Not that it happenned alot, but one self driving vehicle ran down a pedestrian.

You can go onto the Nova website but unfortunately, if you are in Canada, you cannot download the episode.

I'd love to believe that technology can do anything anytime. Mostly a former pilot, not an engineer, but did work for simulator engineering companies and therefore with engineers for short bursts between layoffs. Great and interesting stuff, a whole different world, but I feel that many tend to think that engineering can do anything and anytime. Flight simulators typically have tons of snags out of the box that need to be corrected over time. And that is just a run of the mill simulator. Even an engineer will tell you there are limitations in the tech world.

[Posthumane]

Do you fly IFR to busy places Posthumane? Have you ever experienced the mix of IFR and VFR in places like that such as instructions to “fly the Hudson”? Have you ever actually flown an automated aircraft and progressively scaled it back all the way to full manual flight because it was required? I’m guessing not.

You’re also still focused on manipulating the controls. You’re talking about moving humans up the command chain from one level to another, and failing to realize the pilot already is active and required at all those levels. You mistakingly think you can make strategic decisions that will eliminate tactical ones...you can’t. Saying “no” could be done long before the flight is scheduled to leave, or it could be made in the span of one second when you decide to abandon a landing because of what’s happening, or more importantly what you think will happen. That is just one of countless and unpredictable situations that requires judgement, which requires an active, trained brain.

Has real AI capable of that been invented yet?

First off, your constant attempts to discredit posters by appeal to authority are a red herring. You can assume I'm a 10 year old kid sitting in my parents' basement for all it matters. The question you should ask is do the design teams have this experience? The answer to that is yes. The combined experience of the people involved in these projects is far greater than yours.

You didn't fully understand what i wrote. Humans obviously make decisions at all levels of the command chain in manned aircraft. Its not the strategic decision that would absolve you from having to make tactical ones, its having someone, or some thing, that is reliable making those tactical decisions for you that would allow you to not focus on those. So, have systems bern developed to abort a landing in a split second because of what's happening or what will happen? Yes! That particular task, monitoring progress, conditions, system operability, has absolutely been done by computers. In fact that technology is at a high enough TRL that its been used on aircraft prototypes flown by the end user.

One other thing you utterly fail to consider is the requirement for humans to deal with and decide on human issues. We fly humans around Posthumane, not computers. We frequently deal with human issues and conduct our flights accordingly.

Has real AI capable of that been invented yet?

What human issues specifically? If dealing with passengers is your biggest hangup, then image this whole thing in the context of a cargo flight. Or are those not real airplanes requiring real piloting in your mind?

Engineers work best when they’re supervised by subject matter experts telling them what their product is supposed to do. So if you don’t tell me what it takes to fly airplanes I won’t tell you what it takes to design high level systems.

Its a good thing the engineers on these projects DO work with subject matter experts, and those experts are telling them to keep going. Why they've never consulted you will have to remain as one of those unsolved mysteries I suppose.

[digits_]

Not to mention, before it can be operated in a 703 operation it has to learn to bust minimums, to ignore MEL items, to only snag a plane in a maintenance base and soo much more. You need real pilots to do that!

[Rockie]

First off, your constant attempts to discredit posters by appeal to authority are a red herring.

My intent is not to discredit you, but to make you aware of the giant gaps in your thinking.

You can assume I'm a 10 year old kid sitting in my parents' basement for all it matters.

Not really, I don't know any 10 year old engineers with a pilot's license. Maybe you're just sensitive.

The question you should ask is do the design teams have this experience? The answer to that is yes. The combined experience of the people involved in these projects is far greater than yours.

I'm sure Boeing does have many subject matter experts guiding the engineers in studying this, and the one in charge of this study quoted in the article seems to completely agree with me. Read his quote again.

What human issues specifically? If dealing with passengers is your biggest hangup

In this whole years long discussion I think it's the first time I've brought this particular point up, so where do you get the idea it's my biggest hangup? It is a factor that engineers do not typically consider which is why they need to be closely supervised, and the fact you have to ask what human issues I'm talking about proves my point better than anything I could say myself.

BTW, did you know in the B787 there are many environmental conditions that exceed the autopilot's capabilities and autoland is not approved. There are many equipment faults that seriously degrade the autopilot's function, and by a very quick search of the QRH at least 3 that render the autopilot inoperative completely. I'm sure there are more that Boeing hasn't anticipated or figures will never happen so don't need a procedure for. Airbus used to think like that too when the 320 first came out. Their education has been a hard and bloody one over the years.

First rule when an engineer extols the virtues of his unproven creation...don't believe him.

[mbav8r]

I have to admit that I haven’t read this whole thread so if this has been covered, my apologies.
Posthumane, you mention the technology exist for a split second decision to go around for various scenarios but my thoughts go back to outside forces.
What is or would be the latent delay for a remote operator in a situation where tower issues a go around because someone crossed the hold short line or misses the hold short altogether as one example of a reason to go around, a second or two could be a huge difference in some situations?
Next, a completely infallible autoflight system would have to be designed, completely infallible, how many redundancies would be needed for that? At what point would the cost of the system outweigh the perceived need?
What would the computer do fighting an onboard fire in the avionics bay? This is a situation that requires two pilots and probably worst case scenario, one to hand fly another to run the checklist to fight the fire, would the computer consider of off airport landing or even recognize and be able to implement that plan.
Far too many variables and some smart people will at some point figure they have it covered until a, “we didn’t account for that” situation comes up, for me I hope I don’t see this in my lifetime or that my kids don’t either!
On the topic, ALPA has authorized all means available to stop this initiative, I for one hope they succeed.

[trey kule]

When engineers designed aircraft, I don’t think they ever expected a pilot to stall an aircraft and then pull back on the stick. But the crew in AF did exactly that. And at the end there were three human pilots in the cockpit..
And Indont think engineers ever expected that a herd of senior pilots would do an approach and when they got to low/slow, not think about adding power...but they did.
Or did the engineers anticipate a flight crew identifying and feathering the wrong engine...
But they did....

In the grand scheme of things, there will be accidents. This insane belief that there should be zero accidents is great for getting higher budgets to govt departments, but it is , at least at this juncture, unrealistic.

The question is can one, two, even three flight crew members be eliminated or substituted with a ground based pilot...think a cruise pilot on the ground, one in the cockpit, and one flight crew resting on long hauls.
That technology is not far off, as it does not eliminate the human factors Rockie likes to continually stress.

The issue with this discussion, I see, is twofold. One, the posters who want to go to extremes to prove their point, and 2. Not considering the balance of probability that more technology will be eventually a suitable replacement. Not accident free. But reduced accidents is the goal here. And the issue of passangers is not relevant as I expect it will begin with cargo only, long haul, and reduce the inflight crew from 3 to two.

[Rockie]

When engineers designed aircraft, I don’t think they ever expected a pilot to stall an aircraft and then pull back on the stick. But the crew in AF did exactly that.

Did the engineers expect the pitot tubes to ice up, the air data computers go crazy and the autopilots to fail? If so why didn't they prevent it so the crew wouldn't be put in that position in the first place?

[trey kule]

I expect they will in the future. No learning curve for pilots, right?
And maybe add a canister of gas to neutralize the flight crew and a few lines more of code to cause a pitch down...
or maybe some kind of capusule so the Capt would get 8 hours of alcohol free rest.....never heard of automation staying up all night before a flight....but that is some of the charming emotional human charm you kerp speaking of...

We are never going to agree as I am getting to much enjoyment from disagreeing....

[Rockie]

We are never going to agree...

Well, we agree on that.

[trey kule]

Well played.

[C-GGGQ]

The point of ground based pilots for military drones is to keep people out of warzones and harms way. There is no practical reason for ground based crews on a passenger plane. Its an unecessary complication.

[confusedalot]

So now we are talking ground based pilots on a pilotless aircraft thread. What's the point of putting them behind a desk? You still need a pilot. Who cares where he is situated. You are still paying a salary to that person.

It appears to me that you can take an off the shelf well equipped airliner today and make it operate on it's own with a few more automated systems. All you need is auto taxi, takeoff, collision and weather avoidance, gear and flap actuating, autoflight selections and yer done. All within current technology, actually ten year or more technology. Weather avoidance may be a bit dicey though, weather radar is good but not great, still takes alot of human interpretation. Handling an emergency through automated means is really no big deal....cargo fire...no problem.....fire the bottles, divert to closest airport.

ATC can easily be provided with the tools to direct the machine with a mouse or some very basic keyboard commands, therefore ensuring separation, by replacing verbal or cpdlc type communications. No big deal there.

But, ATC effectively becomes the operator of the aircraft, raising responsability and legal issues.

Even though it can be done in a perfect world, I can't see it happenning anytime soon. Primarily because of cost, but also because of regulatory inertia and, down the road, acceptance by the ordinary person on the street. Met a few non aviation people over time who are absolutely terrified of airplanes.

[Posthumane]

I'm sure Boeing does have many subject matter experts guiding the engineers in studying this, and the one in charge of this study quoted in the article seems to completely agree with me.

Perhaps you should read what he said again. You said the task is impossible until a true ai who is self aware is developed. He said that the technology won't be used until it can deal with emergency situations as well ss human pilots have. They are not at all the same statements. His I agree with.

BTW, did you know in the B787 there are many environmental conditions that exceed the autopilot's capabilities and autoland is not approved. There are many equipment faults that seriously degrade the autopilot's function, and by a very quick search of the QRH at least 3 that render the autopilot inoperative completely. I'm sure there are more that Boeing hasn't anticipated or figures will never happen so don't need a procedure for. Airbus used to think like that too when the 320 first came out. Their education has been a hard and bloody one over the years.

First rule when an engineer extols the virtues of his unproven creation...don't believe him.

Did you know that the 787 is an aircraft designed to be piloted by humans? Like the airbus, although it has some piecemeal automation to aide the pilots, it was obviously never designed to operate without them. Systems like the autopilot are not designed to the practical limit of reliability and redundancy because they don't have to be. Things like powerplants are, at much greater cost, because a total power failure carries much more risk than an autopilot failure. After all, the pilots can't pedal the damn thing. But even thren, total power failure can still happen, and that's simply a risk that's accepted. In autonomous aircraft where the autopilot is a critical item they are made with the same levels of redundancy as the powerplants. In small, cheap aircraft that means one questionable powerplant and one questionable autopilot, and in larger systems that can pose a real hazard it means multiple systems with error monitoring and fault tolerance, which can fly with several control system failures.

Btw, i have yet to see an organization where thr designers and engineers are supervised by tge equipment operators. The SMEs are typically other, more senior engineers.

Did the engineers expect the pitot tubes to ice up, the air data computers go crazy and the autopilots to fail? If so why didn't they prevent it so the crew wouldn't be put in that position in the first place?

Indeed they did. Which is why they made alternate control laws which come into effect when the systems have some failures, relying on the pilot as the backup system. I'm sure you know a lot about alternate laws, since that is actually your area of expertise.

[Rockie]

Still missing the obvious point Posthumane. Dealing with all the possible emergencies requires a self aware real AI. You could cherry pick something and in 50 years build a system to deal with that one narrow set of circumstances, but we are required to deal with everything. You just don’t get that and never will it seems. That’s also why engineers are not supervised by senior engineers, they are supervised by SME’s.

Years ago in a different life my company hired at great expense software designers to build a scheduling program. These people knew how to build software...they did not know anything about what it was actually supposed to do and it was a dismal, expensive failure.

Engineers by themselves do not know anything about physically flying an aircraft and certainly know nothing about commanding one. How could they?

Stay in your lane.

[Posthumane]

Still missing the obvious point Posthumane. Dealing with all the possible emergencies requires a self aware real AI. You could cherry pick something and in 50 years build a system to deal with that one narrow set of circumstances, but we are required to deal with everything. You just don’t get that and never will it seems.

Let me know when a human is built that can handle everything in all circumstances. When they start building those, then all accidents will cease to exist. In the mean time, we will continue to have a non-zero failure rate of both man and machine. Which failures do you know that an autonomous system can't handle and what is your evidence?

That’s also why engineers are not supervised by senior engineers, they are supervised by SME’s.

This has been a trend in all the engineering jobs you've held? I suppose your experience working as an engineer differs from mine as every supervisor I've had was a senior engineer or scientist. They were all SMEs in their fields of course.

Years ago in a different life my company hired at great expense software designers to build a scheduling program. These people knew how to build software...they did not know anything about what it was actually supposed to do and it was a dismal, expensive failure.
[/quote ]
Ah, well that of course settles it. Your company once bought a bad product from a software company and therefore no product can ever be made to work well. Btw, who was in charge of specifying what it was actually supposed to do?

Engineers by themselves do not know anything about physically flying an aircraft and certainly know nothing about commanding one. How could they?

Firstly, some do. Engineers are not a unified group who all have the same knowledge and experience. Second, when a person doesn't know something, they go out and learn about it. That often involves bringing on other people with more exoerience in the subject. I don't work for Boeing, but I'm willing to bet that they have access to some experience with aircraft and flying them.

Stay in your lane.

Sure thing. I'll stick to the threads that talk about building new systems, and engineering, and autonomy, and military, and maybe a little bit of aircraft maintenance. I promise to stay out of threads about air canada union disputes.

[goldeneagle]

Engineers by themselves do not know anything about physically flying an aircraft and certainly know nothing about commanding one. How could they?

Stay in your lane.

I read the rants with interest, would love to comment in detail, but I am contractually bound to not discuss much of this in a public forum.

I am an engineer, my original education was in aircraft design, specifically flight control systems, did that back in the late 70's and early 80's. I did computer options in those days, became a 'recognized expert' in digital software for such systems. I also hold an ATPL and spent more than a decade sitting in the front seat of airplanes, quit that role in '90, and was immediately snapped up by an engineering firm working in this field due to the flying experience combined with the engineering background.

I find your rant interesting, because it's so far from the truth. The team I originally started out with back in the day was fully populated with engineers, all of whom had either commercial or military flying experience. Some were there because they lost a medical, some were there because the preferred the schedule and working conditions, and some were there because they just simply got bored with driving airplanes from point A to point B. To a tee, a couple things we all agreed on, the money was FAR superior to what was available for a cockpit position, and the work is far more interesting. The positions attract the cream of the crop, and folks with only one area of expertise need not apply. To work in senior position that area on the bleeding edge design teams you need expertise in at least two, preferably three disciplines.

I know that folks currently flying the line consider automation unthinkable, but, reality is they will be the last folks consulted on the process. The folks at Boeing and Airbus have plenty of in house expertise in both the design and operational side of this business. I've sat in design requirements meeting with folks from both of those companies and the breadth of experience and knowledge in those meetings is far beyond what you give them credit for. They know how airplanes work, they understand operational requirements, and they understand regulatory requirements. That's why they are in the positions they are.

The writing is already on the wall for the process moving forward. The fate of the second pilot will become the same as the fate of the FE position over the next generation of airliner design cycles. Current generation of production was designed around regulatory requirements for two folks up front, much like older equipment was designed around a requirement for 3. Eventually the FE slot was essentially replaced by the FADEC. The next generation will be very capable of operating with just one and the second seat will be available for an optional crew member. Much like the method used to bring ETOPS online, there will eventually come a proving phase, then it will start to deploy wider, and by that time the single pilot airliner will be the norm for new production, older equipment will continue to fly with two as they were designed. Kinda iike there are still a few older machines flying with 3, but that number is now approaching zero as old airframes are retired.

The generation after that will be designed around the concept of 'pilot optional', and the process will begin again heading into a proving phase. I wont see it in my lifetime, but, my grandchildren (currently 2 years old) will have the opportunity to fly in a fully automated people hauler in the later portion of life. By then, there will still be a few very old 787 and 380 type machines flying around with two folks up front, the majority of the fleet will have one, and the bleeding edge new stuff will be rolling out as 'pilot optional' and you can bet your last dollar some low price carrier will be flying them that way to try grind a few cents off the ticket price.

Feel free to rant all you want, but, in the end it's essentially the horseless carriage debate starting anew. In time things will change, like it or not, it will happen.

[DrSpaceman]

Creating a wireless link to operate an aircraft imo is completely irresponsible. To go completely pilotless you need true AI, and at that point there are WAY more jobs that can be automated for way more savings than pilots. It’ll create a paradigm shift too big for us to comprehend today.

[Rockie]

Once again I will state that it’s not impossible, but it is not possible without real AI. Boeing agrees.

Boeing Co (BA.N) is looking ahead to a brave new world where jetliners fly without pilots and aims to test some of the technology next year, the world's biggest plane maker said in a briefing ahead of the Paris Airshow.

The idea may seem far-fetched but with self-flying drones available for less than $1,000, "the basic building blocks of the technology clearly are available," said Mike Sinnett, Boeing's vice president of product development.

Jetliners can already take off, cruise and land using their onboard flight computers and the number of pilots on a standard passenger plane has dropped to two from three over the years.
Sinnett, a pilot himself, plans to test the technology in a cockpit simulator this summer and "fly on an airplane next year some artificial intelligence that makes decisions that pilots would make", he said.

Self-flying aircraft would need to meet the safety standards of air travel, which had its safest year in 2016, according to the Aviation Safety Network. They would also need to convince regulators who don't yet know how to certify such planes.

"I have no idea how we're going to do that," Sinnett said. "But we're studying it right now and we're developing those algorithms."

Airlines are among those backing the idea, in part to deal with a projected need for 1.5 million pilots over the next 20 years as global demand for air travel continues to grow.
But a self-flying plane would need to be able land safely as Captain Chesley Sullenberger did in the "Miracle on the Hudson," Sinnett said. "If it can't, then we can't go there."

What do you think is required to make the radically out of the box determination that a river landing is your best chance of survival? What about figuring out how to fly a DC-10 with no hydraulics? What about when the power goes out completely? How about an out of control cabin fire or running out of gas?

You think an Ipad’s going to make the right decisions?

Real artificial intelligence. When you have that let me know...

[mixturerich]

Guys, listen. We may one day have pilotless airplanes, but that day will be a very long time from now. We barely even have automated freight trains, buses, or cargo ships. They haven’t removed the crew from those, so why would they remove the crew from airplanes. Besides, the new aircraft have 40+ year service lives and I don’t foresee it being cost-effective to retrofit one to become pilotless. So for now, and for probably half a century, we can all just chill out.