"Looking Forward" to a Pilotless Future

[Ian W]

I thought I would put this here as it deserves some technical discussion. The thread title comes from a Bloomberg article:

"Airbus Looking Forward to a Pilotless Future"
Airbus SE (https://www.bloomberg.com/quote/AIR:FP) is looking to develop autonomous aircraft and technologies that will allow a single pilot to operate commercial jetliners, helping cut costs for carriers, Chief Technology Officer Paul Eremenko said.“The more disruptive approach is to say maybe we can reduce the crew needs for our future aircraft,” Eremenko told Bloomberg Television’s Yvonne Man in an interview broadcast on Wednesday. “We’re pursuing single-pilot operation as a potential option and a lot of the technologies needed to make that happen has also put us on the path towards unpiloted operation.”
https://www.bloomberg.com/news/articles/2017-11-21/airbus-exploring-single-pilot-autonomous-aircraft-in-tech-race

Read the article before continuing.

The bean counters are already convinced, the technical aspects are being solved mainly in military operations and in smaller commercial 'drones' where attrition is more acceptable.

Large military manufacturers have been and are successfully demonstrating multiple in flight refueling and carrier landings - both technically extremely difficult. Similarly, military aircraft manufacturers have demonstrated adaptive flight control systems that can provide apparently normal aircraft handling even after quite dramatic battle damage. Against these simple passenger or cargo flights even with the occasional emergency seem to be relatively straightforward to automation.

So the question is are there technical and/or safety reasons that can be provided that are insurmountable to plans for what seems to be the decided progression.

I don't believe that the emotional and attitudinal arguments will have much weight against the financial ones. We are in an era of driverless trains and driverless cars - yet I can remember when elevators were considered unsafe without an operator - so I suspect that people could in future accept pilotless passenger aircraft.

[Jurow]

To add fuel to this reflexion with Boeing view on this evolution.

https://www.seattletimes.com/business/boeing-aerospace/boeing-studies-planes-without-pilots-plans-experiments-next-year/

The emotional impact of a plane crash is massively more important than a car or bus crash.

The fear of being flown by an unmanned system could have direct consequences on the number of passengers travelling. Without passengers, no profit.

[vilas]

When the industry saves 35billion USD annually the pilotless aircraft can afford to have better redundancies obviating any real loss of services through failures. So initially single pilot for passenger confidence then subsequently no pilot operations. The writing is on the wall.

[parabellum]

There are several threads here on PPRuNe discussing the relative merits, or otherwise, of a pilotless commercial passenger jet, the subject has been done to death.


1. Until terrorism is totally under control, if not wiped out, remotely controlled pax aircraft won't happen. A ground based remotely controlled system will need to be 100% immune from any form of electronic 'hacking' and any form of suicidal terrorist attack.


2. Without 1., (above), the international insurance market won't touch it, no insurance = no fly.


3. The cost of overcoming 1 and 2 (above) - probably insurmountable and therefore not commercially viable within our lifetimes.


4. No one has presented verifiable figures on the cost benefit of a totally remotely controlled system versus what we have today. Simply getting rid of airborne pilots, in the great big scheme of things, may not be commercially viable at all. Fifteen hour sectors, one pilot? I don't think so.


5. Repeat 1. (above).

[Jeffory]

Surely ATC would be another hurdle unless every other aircraft is also pilotless?

I'm thinking there will be a lot of unforseen job creation in order to service/maintain, ensure certification/compliance of systems, troubleshoot errors and deal with the excessive amount of new red tape that will come with completely pilotless aircraft. Have the cost savings been fully understood?

Single pilot could be all it comes to in 20 years time, and it will be trialled in cargo operations first no doubt.

Driverless trains and cars can roll to a stop or have a low speed collision (best case scenario). But hey, maybe ballistic parachutes will be attached to large jets to allow a "safe" way to crash ;)

This is still in it's infancy

[megan]

QF32, how does Airbus figure how a computer would handle that? Pilotless airliners? Not for this little black duck, nor single pilot.

[Atlas Shrugged]

Yes Megan. Absolutely!

There's always a some sort of group demanding pilots be replaced with automation and the problem with that is that there are numerous events every day, although not entirely like QF32, where pilots save the aircraft from the effects of automation...but, a save is just part of a normal day.

The biggest problem is with partial automation - by having a pilot there to take over when the system $hits itself, but who otherwise sits and does nothing for hours on end. In other words, he's only there for when the going gets really tough, but the automation won't keep him in the loop nor will it keep his skills in practice. Unrealistically, it will expect him to go from brain dead to top of his game in an instant.

'Complete' automation may happen one day but, they will need to get it to the point where the software is 100% perfect, AND the programmers will have thought of EVERYTHING, or at the very least come up with an AI that can handle EVERYTHING. As they can't even keep my desktop computer or my iPhone on line properly, I imagine they've got quite some distance to go.

If I'm still around when it does happen, you will more than likely find me running very quickly in the opposite direction......

[OK4Wire]

KayPam,

I have recently heard that very same issue being confirmed to me by a colleague returning from some time with airbus. Interesting.

[Smooth Airperator]

How about replacing CEOs with algorithms that have the long term goal of airlines in mind and not short range bonuses?

[Ian W]

There are several threads here on PPRuNe discussing the relative merits, or otherwise, of a pilotless commercial passenger jet, the subject has been done to death.


1. Until terrorism is totally under control, if not wiped out, remotely controlled pax aircraft won't happen. A ground based remotely controlled system will need to be 100% immune from any form of electronic 'hacking' and any form of suicidal terrorist attack.


2. Without 1., (above), the international insurance market won't touch it, no insurance = no fly.


3. The cost of overcoming 1 and 2 (above) - probably insurmountable and therefore not commercially viable within our lifetimes.


4. No one has presented verifiable figures on the cost benefit of a totally remotely controlled system versus what we have today. Simply getting rid of airborne pilots, in the great big scheme of things, may not be commercially viable at all. Fifteen hour sectors, one pilot? I don't think so.


5. Repeat 1. (above).

The word used by both Airbus and Boeing was autonomous. This is not a 'remotely piloted aircraft system' RPAS. It is a fully automated autonomous aircraft. That of course does not mean that there will be no issues with insurance but the savings of an autonomous system could allow insurance rates to go a lot higher and still make the autonomous aircraft significantly cheaper.

The human on the loop systems in most of the modern aircraft are effectively autonomous aircraft with a pilot on standby watching in theory ready to pick up the bag of bolts if something goes wrong. The autonomous systems planned would be more capable than the current systems and will not have the easy escape for the system designer of "if this happens give it to the pilot".

[Uplinker]

How about replacing CEOs with algorithms that have the long term goal of airlines in mind and not short range bonuses?

Well said. If only.
:ok:

[dual land]

The human on the loop systems in most of the modern aircraft are effectively autonomous aircraft with a pilot on standby watching in theory ready to pick up the bag of bolts if something goes wrong.

I wouldn't quite describe them as such. Pilots make a large amount of decisions and inputs even if they are using the autopilot.

[fantom]

We all know the one about the pilot and a dog.

How will the public take to the pilot and a battery?

[737er]

I love it when all the clueless nerds start talking about this subject. It not even worth the energy. None of them alive now will live to see a pilotless pax airliner, even if they are kids.

[parabellum]

Pont taken Ian W, so just the terrorist threat left to deal with? ;)

[neville_nobody]

I have seen discussions on this in other forums and basically it is a non starter.

Whilst pilot less aircraft individually are technically possible now, a automated airliner in a automated aviation system is just not going to happen any time soon, if ever. There are just so many unintended consequences and problems evolving from that it is bordering on impossible not to mention less safe than the system we have now. The other issue of a pilotless system is just the massive infrastructure build to make it work.

All that the tech nerds are doing is replacing one problem, 'pilot error', with a whole bunch of new problems that have not been tested yet, additional to other new problems that they haven't even considered. The old 'unknown unknowns'.

Single pilot is easy to accommodate as it already happens now, whether it is a good idea or not is a another question all together.

[Ian W]

"When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong."If an elderly but distinguished scientist says that something is possible, he is almost certainly right; but if he says that it is impossible, he is very probably wrong.

Arthur C Clarke.

Autonomous aircraft are already flying and have been for some years. Plans are well advanced for pax carrying autonomous aircraft as air taxis - so we will see if there are queues for the service or no takers.

Dubai Is Test-Flying Its Future Air Taxis (http://www.popularmechanics.com/technology/infrastructure/news/a28390/dubai-begins-air-taxi-tests/)

https://www.uasvision.com/2017/11/09/uber-announces-unmanned-air-taxi-deal-with-nasa/

https://www.theguardian.com/technology/2017/nov/08/uber-signs-contract-nasa-develop-flying-taxi-software

[73qanda]

When the industry saves 35billion USD annually the pilotless aircraft can afford to have better redundancies obviating any real loss of services through failures
The problem with that theory is the cut throat nature of the industry. There won’t suddenly be $35B in extra profit shared about the global industry, the price of tickets would drop maintaining the thin margins and the pax would get to their holiday destination for $65 instead of $72.
An increase in safety is the only real benefit I could see ( assuming there is an increase).

[Marlon Brando]

Let's do some maths.

Captain in low cost european airline makes 10000€ plus social, pension..., if any.
Let's say it costs 16000€ to his companie a month.
Copilot costs 10000€
Total 26000€ .

These 2 guys fly 15 days a month, 4 legs, 200 pax every flights.
They fly 15x4x200 = 12000 pax a month

26000€/12000 pax = 2.17 € per passenger.
For 2 pilots.

You rather buy a ticket 100€ with no pilot, or 102.17€ with 2 ?

[KayPam]

The word used by both Airbus and Boeing was autonomous. This is not a 'remotely piloted aircraft system' RPAS. It is a fully automated autonomous aircraft. That of course does not mean that there will be no issues with insurance but the savings of an autonomous system could allow insurance rates to go a lot higher and still make the autonomous aircraft significantly cheaper.

The human on the loop systems in most of the modern aircraft are effectively autonomous aircraft with a pilot on standby watching in theory ready to pick up the bag of bolts if something goes wrong. The autonomous systems planned would be more capable than the current systems and will not have the easy escape for the system designer of "if this happens give it to the pilot".
Your reasoning has a bias.
Why would the insurance rates go up ? Because airplanes would crash more.
Therefore it would be inefficient, since passengers care so much about safety (or rather, since passengers overestimate so much the danger of flying)

[WindSheer]

If you are all too niave to believe that pilotless aircraft will begin with cargo, followed by many years of review and safety validation and followed eventually by pax pilotless then......or wait a minute, am I too niave?

20 years ago we hardly had Internet, let alone all the recent introductions of smart phones etc.

Another 20 years will bring about enormous evolution, unless of course it goes too far and gives 'some of the more idiotic' leaders of the world the avenue to bring about our own destruction.

While there are cargo planes or trains out there, they are the perfect avenue to test new concepts.

Just being real.

[Ian W]

Your reasoning has a bias.
Why would the insurance rates go up ? Because airplanes would crash more.
Therefore it would be inefficient, since passengers care so much about safety (or rather, since passengers overestimate so much the danger of flying)

It may be that the autonomous aircraft are safer - however, the perception (shown here) is that they will be less safe. It is just a perception without supporting evidence. Nevertheless, initially I would expect insurance rates could reflect the perceived risk until there was sufficient history to demonstrate whether the perception was soundly based or not. If the savings to the operators are significant a rise in rates may not be a disincentive.

[tdracer]

Perceptions change. Wait until driverless cars are perfected and the automotive accident rate plummets. Once that happens, every time there is an aircraft disaster caused by "pilot error" there will be an outcry about why do we still have fallible humans piloting these things.
It won't happen soon, but it will happen.
In my lifetime, we've gone from a world where they sold flight life insurance in the airports (because many insurance policy's specifically excluded aircraft accidents) to a world where it is no exaggeration that the most dangerous part of flying is the drive to the airport. 100 years ago most people wouldn't even dream of using an aircraft to cross an ocean - now not only is it common, it's pretty much the only option. Assuming we avoid blowing ourselves up, life 100 years in the future will be similarly unimaginable.
Computers are getting exponentially more capable - humans not so much.

[msbbarratt]

Don't be fooled; computers are getting better only because the humans who program them are getting better. But they're still human, and prone to human mistakes and foibles. I know, it's my job to do that, and frankly I sweat buckets over systems that have to be good but not immediately safety critical.

Moral Question

It basically comes down to this: is it moral to risk a backward step in aviation safety for the sake of some financiers' wet dream of an investment? Especially as the aviation industry is so very good these days at not crashing, having learned from crashes over the past century? There's not a lot of room for improvement in safety, so that's not really a viable justification...

Are we really going to say "let's go through all that again" for the sake of pilotless aircraft?

(Let's not distinguish between cargo and passenger aircraft; people on the ground are just as dead if either crashes into their house. Incidentally one of the costs of UAV testing is finding somewhere where there's no one to crash land on top of...)

I refuse to participate in any effort to develop software for an automated airliner, or car (i.e. something intended to operate constantly in or over a human-dense environment. Most UAVs are operated in sparsely populated areas, so am less concerned about these at the moment). Trains are different; they operate in an environment that is almost entirely under our control if we choose so (i.e. avoid rivers, cliffs, etc). That means that we can write software to control them autonomously; we really can think of almost everything they have to do. They also benefit from being much bigger and heavier that most of the movable things that might get in their way, and they don't go wondering away from their rails very often...

Why it Happens

There's what is called a "conspiracy of optimism" surrounding AI and transport at the moment, and those have a nasty habit of being self perpetuating until some unsuspecting user pays the price of the thing being not as good as it needed to be.

There's a ton of investors out there who are putting huge sums of money into self driving this, self flying that, on a "just in case" basis. It doesn't take much of a sales pitch from a bunch of enthusiastic engineers to convince some very big wallets to splash the cash.

The problems arise when some return on that investment is demanded. The technology is very obviously not up to expectations, and so the autonomous car industry tries to make it a political issue. There are PR campaigns running; effectively some are seeking to have the "rules of the road" changed to suit the industry; that will be at the expense of all road users.

It is the same with self flying passenger aircraft; the idea of having them is rapidly transforming into a self perpetuating "we must have them", without any real explanation of why, just like it has with cars. There's lots of dubious justifications...

In the aviation industry we already have rules, engineering standards, etc. for certification of software and hardware for safe flight. These have been very effective at putting aviation safety at a very high level.

The trouble is that I cannot see us ever successfully developing an autonomous pilotless aircraft whilst sticking to the existing rules, certification procedures, etc. It'd would be enormously difficult to do. What I can see happening is various companies, backed by very rich investors, applying pressure to have the rules relaxed to make it easier for their technology to be put into the sky. The rule book is in the way? Burn the rule book. Yet it's these rules that have made the industry as safe as it currently is.

And because we cannot build one that's guaranteed to be 100% reliable the only way we're going to find out if they're worthwhile from a safety point of view is to build lots of them, fly them with paying passengers and let the crash stats build up over decades whilst the elements, terrorists, hackers, maintenance crews and airline management do their worst.

That is, use the paying public as guinea pigs, yet again.

And for all that time we're just one software bug or undreamt of hardware failure or network security issue or hacker away from killing a lot of people. That's for only the sake of what is demonstrably a very marginal improvement for passengers and a very doubtful financial gain for the airlines and manufacturers.

And it might truly be a lot of people who end up dying to prove the point; what if one ATC net got hacked and issued convergent directions to 1000 airborne aircraft at the same instant?

The cost / benefit / risk analysis is poor in my estimation. I wouldn't want to be partly responsible.

Paranoia or History Repeating?

Ok so perhaps I'm being paranoid. But humans are very poor at engaging rationally with the prospect of risks that have huge consequences that are probably fairly unlikely to occur. TEPCO didn't and look where that's got them.

Humans are also very poor at adding risks to the equation in the first place. For example how do you assess the risk of being hacked? Can't do that, so let's not consider it and assume the security will be good enough (that's the usual response). Yet it's 100% guaranteed that hackers will try... Just like it's 100% guaranteed that thieves will hijack driverless lorries.

Vanity

Vanity is a dangerous thing in this business, and it's present in the self driving car industry. To illustrate the "problem" in the self driving car industry, consider the possibility of autonomous cars being "bullied" by humans (they won't drive into me, so I can intimidate it!). When asked on BBC Radio recently (Tech Tent, 10th Nov 2017), an industry personality was deadly serious about solving this problem with laws. Seriously? It becomes illegal for you to act in a way that is interpreted as a danger by someone else's lame brained self driving car? No way! How vain is that, expecting everyone else to be compelled by law to account for the nature of one's own product!!!

Airports and Everyone Else

I don't think the airports would value a self-flying, pilotless autonomous airliner. Want to take some landing aid out of service? Why no you can't, the planes can't land without it. So you'd have to have two, just in case.

The knock on costs to other players are going to be quite large, and that will simply get passed on to the paying passenger; manned aircraft would probably be cheaper to land.

[KayPam]

There are foundamental differences between cars and planes.

Airliners are operated by very few professionals.
Cars can be operated by almost anyone over 18.

Basically, a driver has the liberty of driving like an idiot all he wants. You can drive several years without seeing a single policeman (and handing him your papers).
So obviously, the accident rate would plummet if you had self driving cars. But it would also plummet in a very similar fashion if all cars were operated with the same standards as airliners are : 2 drivers with 200 hours of practical training, years of experience, an operating manual to rule everything, regular and exhausting training in the sim, an other crew member to monitor what you're doing and pax to complain of the slightest problems...

The only question is : do we need self driving cars ? No, driving is not terrifying, it's actually already very safe.
Do we need self driving planes ? No, it's already even safer.

[msbbarratt]

There are foundamental differences between cars and planes.

Airliners are operated by very few professionals.
Cars can be operated by almost anyone over 18.

There are some similarities; both are difficult challenges for full autonomy (cars probably more so - roads are an even more uncontrolled environment that the air), and the kind of people most keen on achieving it aren't necessarily the one's I'd trust to make a fully objective assessment as to the wisdom of attempting it. Especially as they're commonly the ones also keen on eroding the normal rules and processes surrounding safety-critical systems development, testing and certification.

Basically what I think will result from the current AI-Transport bubble is a bunch of "partial" projects


Cars that are self driving some of the time, in some circumstances.

But what's the use in that? If it can't bring me home from the pub half cut or take the kids to school, I'm not interested.


Airliners that might allow pilots to nod off for a period of time.

So not so different from today's tech.
How does anyone think that the world's aviation infrastructure can be updated and certified to the point where we trust an airliner to 1) land at any airport on route / diversion, 2) taxi to the right place, 3) do all that in the foulest and fairest of conditions, 4) do all that with the requisite ground infrastructure going offline just as it commits to a landing, all this being basically the only thing that a modern airliner isn't trusted to do for itself today?
I'd be interested to know exactly how an autonomous airliner is supposed to declare a Mayday or Pan, and what exactly is supposed to happen when it does...

[vilas]

Airlines is just another business like any other. Economics and technological advances will overtake everything and everyone. Only question is when. We discuss so many accidents where people sitting in front have not given the impression of a highly skilled professional. It may be difficult to visualize presently but will happen.

[KayPam]

As someone calculated earlier, front seat crew account for about 2% of the price of a ticket.
Probably even less for long-haul operations (with more pax aboard).

All the sensors and computers required, in addition to the advanced AI required to make the airplane autonomous, will surely cost more than pilots for a very long time.

And by the time it's not the case anymore, it's possible that there won't be enough petrol to continue flying airplanes.
So there's a serious chance we'll never see autonomous airplanes in our life.

[Pugilistic Animus]

What about situations such as a forgotten chock preventing right gear extension...2 go arounds and finally anti skid off to intentionally burst the tires.

Can a computer deal with that?

[EI_DVM]

Out of curiosity, are the latest generation of aircraft (A350, B787, CS300/100) authorised to conduct an autoland with braking action less than good or on contaminated runways, or with turbulance greater than moderate? Just curious as current/previous generation aircraft (A320/A330/B737) are not allowed to conduct autoland ops in these conditions so just wondering if today's modern technology allows this type of landing, or if that's still another generation away?

[Lookleft]

Boeing and Airbus are working on the next generation of narrow body aircraft. They would be at least 10 to 15 years away from service in the most optimistic time frame. They are not being developed as pilotless aircraft and they will be in service for at least 40 years in their various forms. Any attempt to reconfigure them to pilotless will require recertification. If you don't think thats an issue then look at the 737 flightdeck, an aircraft that has been in service for 50 years.

[Sorry Dog]

Vanity

Vanity is a dangerous thing in this business, and it's present in the self driving car industry. To illustrate the "problem" in the self driving car industry, consider the possibility of autonomous cars being "bullied" by humans (they won't drive into me, so I can intimidate it!). When asked on BBC Radio recently (Tech Tent, 10th Nov 2017), an industry personality was deadly serious about solving this problem with laws. Seriously? It becomes illegal for you to act in a way that is interpreted as a danger by someone else's lame brained self driving car? No way! How vain is that, expecting everyone else to be compelled by law to account for the nature of one's own product!!!

Airports and Everyone Else

I don't think the airports would value a self-flying, pilotless autonomous airliner. Want to take some landing aid out of service? Why no you can't, the planes can't land without it. So you'd have to have two, just in case.

The knock on costs to other players are going to be quite large, and that will simply get passed on to the paying passenger; manned aircraft would probably be cheaper to land.

You seem to have more foresight of this issue than most.

The thing is the cost/benefit analysis for cars is different than for airplanes. First of all the economies of scale are more favorable for autos. Second there are more potential benefits to society such as access to transport for those that formerly had difficulty operating or using personal autos. Of course, this has me way more fearful of autonomous car parade trampling the driving rights of everyone else. You already gave one example but there are several other foreseeable consequences as well. Traffic congestion is another big one in my opinion. The autonomous fans tend to tout that traffic will be better since the AI drivers are better and more orderly. That may be so, but I highly suspect the overall effect will be much more gridlock. Once cars are untethered from their drivers, the utilization rate will increase meaning the car will be on the road more. Often for trivial benefits of the owner such as have the car circle block endless to avoid parking or I also foresee an explosion of passengerless vehicles performing delivery services. I really believe that it going to be a huge problem in urban and surburban areas and backlash will form as well.

Fortunately, for aircraft, I don't see it being a big issue in years than I have left to fly.

[msbbarratt]

I hope the rest of your flying time has smooth air and comfortable runways, and for many years beyond!

It's quite interesting studying some of the claims the autonomous vehicle people make; safety, traffic improvements, etc. There's also what's being touted as a virtuous circle between electric cars and self-driving cars; you need electric cars to save the environment, you need self-driving to make electric cars viable, ergo please invest megabucks into self-driving cars.

The trouble is that absolutely none of these claims will ever come to fruition whilst being 100% "safe" with the roads and other road users we have today. The danger is that giga-bucks will be poured into changing the roads, banishing other road users, simply to fulfil something that is little better than a religious prophecy. Such expenditure would be an acknowledgement that the original prophecy was unfulfillable, and that the most important thing had become to fulfil a heavily amended form of it even if that is in itself now not worthwhile.

Ask yourselves how many times governments have spent lots of good money on top of bad chasing an already dead and pointless idea, simply because they’ve already started…

Seen it All Before

The whole pattern of research, argument, persuasion, policy, law, and eventual retraction is an oft-repeated human thing. Look at the food industry; for decades the industry, experts, governments, magazines, etc. have been saying "butter is bad for you, it'll make you fat, give you heart disease, cancer, and make you go bald, cause flatulence". A lot of businesses made a lot of money out of trans-fats and artificial substitutes for decades. And then (quite recently) it turns out that the "advice" is rubbish, butter and the like is fine (even necessary) in moderation, cue the butter shortage in France and elsewhere as suddenly the demand goes through the roof.

How many health £ have been spent and how many people have died due to trans-fats? Probably an incalculably large number.

The Power of Lobby

Where there's money to be made, there's almost no end to the measures, arguments, research, advertising, lobbying, voting, etc. that will be wheeled out to create a market. That is exactly what Tesla, Google, Amazon, Uber, etc. are doing right now. Just like the smoking industry did, the food industry, the car industry back in the bad old days, the drinks industry today, the oil industry, etc.

The danger is that a lot of harm will be done along the way from here to there when we start seeing the poor accident statistics / traffic flow figures rolling in. The industry's end goal is to make it too late to go back before everyone has a chance to work out whether it's a bad idea or not.

With self driving cars, what I think will actually happen is that people will not value the inevitable "partial" solution enough for the market to take off, and it'll simply be too expensive to give it away for free.

The only company doing it properly is Volvo - they're saying they'll take the rap for collisions and accidents caused by their software. Everyone else is saying "Read the End User License Agreement".

Musk

Tesla are particularly bad because there's a lot of "Tech Rock God" going on with Musk which causes a severe and possibly harmful reality distortion field. People believe him literally as an article of faith (I know some who are like this). Someone killed themselves misusing Autopilot, yet Tesla are not even trying to nor claim to do a fully autonomous self-driving car (apart from their refusal thus far to drop the term “Autopilot”).

Musk in particular is interesting; Tesla does not make money, and they're only now finding out just how hard it is to mass produce cars; SpaceX does not make money; so what’s in it for him? His personal reputation is his way of making him money, and his co-investors simply have to hope that somehow Tesla and SpaceX will actually turn a profit (but for him that's merely a bonus). Tesla won’t, they’re building the wrong sort of battery (Toyota have the right battery, solid-electrolyte lithium ion, watch this space). SpaceX are unlikely to as they’re trying to make money in an expensive market where margins are already pretty thin, and they’ve wasted a lot through careless engineering brought about by cost and time pressures. Meanwhile their competitors (e.g. Ariane) are fearsomely good at being on time (my own experience of them is to the second, planned years in advance), on cost, and reliable.

People such as this can be very influential beyond their actual achievment, qualifications, or field of expertise. If Musk were to turn his gaze on the aviation industry, things may end up happening that you pilots would have a lot of deep concerns over. You just kinda have to hope that the barriers to achieving anything at all in aviation are too high for even someone like Musk to consider leaping. It depends on what he wants to do. Battery-hybrid aircraft - OK no problems with that. Tesla inspired pilotless aircraft - anyone want to get on board?

[tdracer]

MSB, you make some very good points (some of your comments on Tesla are spot on - as another automotive CEO put it 'it's really easy to build good cars when you're losing money on every one you build').
But you're still thinking "short term" (and by that I'm talking 20-50 years). Like you, I'm involved in aviation software (or more accurately was involved - I retired about a year ago :ok:). I witnessed massive changes in the 40 years I spent in the industry before that retirement. I clearly recall pilots stating they'd "NEVER" get on an aircraft with "plastic" (composite) wings, computer flight controls (Fly By Wire), or electronically controlled engines (FADEC). Being an engine guy, I'm particularly familiar with the reservations regarding FADEC engines, and some of the concerns I heard were quite simply laughable...
Yet as you certainly know, all these things are now commonplace. Software coding today bears almost zero resemblance to what we did in the 1980's - "coding errors" as such basically no longer exist since the coding is done by computers - the weak link now is in the requirements and how they are translated into flow diagrams.
Basically, anything that has ever happened can be accounted for (yes, Pugilistic Animus, that includes a failed gear extension), the weakness being the 'unknown unknown - it's really hard to incorporate human ingenuity into a computer program given today's technology (the often mentioned Sully's Miracle on the Hudson is actually pretty trivial - likely with an even better outcome of making it to an airfield - since it would take a computer milliseconds to determine what Sully did in roughly 20 seconds)(that's not to detract from what Sully did in any way - by human standards he was exceptional).
But that's todays technology. With computer capabilities increasing exponentially via Moore's law (comment - Moore's law isn't a law - it's an observation - but it's an observation that has held remarkably accurate for several decades), can any of us even imagine what computers will be capable of 50 years from now?
As I mentioned, it won't happen soon - likely not in my lifetime. But we're already in a world where pilot error, CFIT, and suicidal pilots result in far more aircraft accidents than mechanical failure. The major auto manufacturers are prediction fully autonomous cars within five years - even if they are off by ten years, it's the future of transportation. When autonomous cars become perfected and commonplace and driving to the airport is no longer the most dangerous part of flying the public tolerance for pilot error accidents will vanish (not that it's ever been very high).

[msbbarratt]

Hello tdracer, I envy you, I'm still slaving away at the keyboard...

The requirements are indeed the weak link. It's possible to completely set out the requirements for a wing, FADEC, fly by wire. We can't do that with machine learning and AI.

And if we can't do that, then the only thing to do is to put them into service and see how many crash over the following decades. Given the excellent state of aviation safety, there's precious little room for improvement.

That's the moral question, and I think that we shouldn't entertain that at all, not simply for the sake of eliminating a company cost.

The fear I have is that the money power will try and brush the morals under the carpet (just as they are trying to do in the self driving car industry).

However if someone can prove in advance of the first revenue flight that it will be a safety improvement (a truly narrow window of opportunity) that would be harder to argue against. But that's impossible to do.

I work with a bunch of guys who do machine learning algorithms, and the one permanent fact that remains is that they need training data. And if that's incomplete, or just plain wrong, you end up with a junk machine. Moores observation doesn't come into it; more transistors does not make our training data set complete, or right. And there's precious little sign of these things exhibitting inspired imagination or adaptability like Sully did.

[Elephant and Castle]

But we're already in a world where pilot error, CFIT, and suicidal pilots result in far more aircraft accidents than mechanical failure.

As the pilot is the last line of defence the failure of the last line on occasion can and does lead to an accident. That line of the defence (the pilot) in the other hand does catch many hardware and software failures and thus prevents many accidents. Accidents that are prevented are not reported. The order of magnitude must be in the order of hundreds of thousands of prevented accidents to caused accidents in favour of the pilot.

It is clear that while operating in the middle of the envelope computers can do a better job than humans. The issue arises when external conditions (WX), procedural or mechanical failures place you at the edge (or beyond) of the envelope. In the case of cars the software defaults to brakes on and STOP. In the case of aircraft it defaults to automation disconnect - manual handling. The difference is significant, the autonomous aircraft will have to be designed with a much greater envelope capability than current aircraft and will require greater system redundancy and a larger number of sensor and sensor types with the added cost that will bring.

The scale of the problem can be seen by looking at FMC irregularities published by both Airbus and Boeing. These are known software glitches that have undesired consequences on the behaviour of the aircraft. In very mature products such as the A320 family or the B737 these irregularities are many pages long. How long then to develop a way more complex system that has no such irregularities at all.

Anyone that flies an Airbus knows that resetting a computer is a daily occurrence to restore normal function. That being the current state of affairs I cannot see a system with the level of reliability required any time soon.

Time is long so no doubt in the end it will happen but its certainly not around the corner.

[mrfox]

from last week:
The complex failure scenario of the second Soyuz rocket launch from Vostochny continued emerging in the days following the accident. Although the culprit had quickly been pinned down by flight control specialists, even seasoned space engineers, who were not directly involved in the intricacies of guidance systems, struggled to fully comprehend the bizarre nature of the accident.

In the Soyuz/Fregat launch vehicle, the first three booster stages of the rocket and the Fregat upper stage have their two separate guidance systems controlled by their own gyroscopic platforms. The guidance reference axis used by the gyroscopes on the Soyuz and on the Fregat had a 10-degree difference. The angle of a roll maneuver for rockets lifting off from Baikonur, Plesetsk and Kourou, which was required to guide them into a correct azimuth of ascent, normally laid within a range from positive 140 to negative 140 degrees. To bring the gyroscopic guidance system into a position matching the azimuth of the launch, its main platform has to be rotated into a zero-degree position via a shortest possible route. The ill-fated launch from Vostochny required a roll maneuver of around 174 degrees (which was apparently conducted from the 5th to 22nd second of the flight), and with an additional 10 degrees for the Fregat's reference axis, it meant that its gyro platform had to turn around 184 degrees in order to reach the required "zero" position.

In the Soyuz rocket, the gyro platform normally rotated from 174 degrees back to a zero position, providing the correct guidance. However on the Fregat, the shortest path for its platform to a zero-degree position was to increase its angle from 184 to 360 degrees. Essentially, the platform came to the same position, but this is not how the software in the main flight control computer on the Fregat interpreted the situation. Instead, the computer decided that the spacecraft had been 360 degrees off target and dutifully commanded its thrusters to fire to turn it around to the required zero-degree position. After the nearly 60-degree turn at a rate of around one degree per second, the Fregat began a preprogrammed trajectory correction maneuver with its main engine. Unfortunately, the spacecraft was in a wrong attitude and, as a result, the engine was fired in a wrong direction.

Soyuz fails to deliver 19 satellites from Vostochny (http://russianspaceweb.com/meteor-m2-1.html#1201)

[msbbarratt]

Accidents that are prevented are not reported. The order of magnitude must be in the order of hundreds of thousands of prevented accidents to caused accidents in favour of the pilot.

I think it would be interesting and illuminating if pilots themselves organised collection and reporting of such data, independent of their companies. It would serve as a good measure of how important pilots are.

In the case of cars the software defaults to brakes on and STOP.

Probably, yes. Unfortunately there are occassions when stopping would be the wrong thing to do!

The scale of the problem can be seen by looking at FMC irregularities published by both Airbus and Boeing. These are known software glitches that have undesired consequences on the behaviour of the aircraft. In very mature products such as the A320 family or the B737 these irregularities are many pages long. How long then to develop a way more complex system that has no such irregularities at all.

Glitches and irregularties are OK so long as they're known about, can be worked around, at which point they become "quirks" (a technical term...). The danger within an AI machine learning based system is that the number, severity and exact behaviour of quirks is not quantifiable, even after long operation; the first thing you may know about it is when you look out the window and wonder why the ground seems to be looming large...

The autonomous car industry is in effect hoping that it never has to prove that their systems "work and are an improvement on humans in all circumstance" in advance of them going into mass production.

Anyone that flies an Airbus knows that resetting a computer is a daily occurrence to restore normal function. That being the current state of affairs I cannot see a system with the level of reliability required any time soon.

Time is long so no doubt in the end it will happen but its certainly not around the corner.

I can't see it happening in the end, not with the state of technology we have now. With today's machine learning / AI systems we cannot say exactly what it is we have built; thus it cannot be certified, "examined", etc. Too dumb to be fully trusted, too clever for their behaviour to be fully analysed and understood. Not a good combination.

To really get there we'd need AI more or less as portrayed in Sci Fi films like I, Robot (we best hope we don't end up with Marvin the Paranoid Android). And that's a looooooong way off. In fact we haven't the first scoobies of an idea how to actually, really, do that.

At the risk of going down a deep rabbit hole, Roger Penrose (mathematician) has written some intersting observations on how the brain works. The Turing Machine Halting Problem is interesting; a Turing machine cannot tell you that another Turing machine will complete its program without running that program (except in trivial cases). Yet a human brain can look at a program and work it out. Penrose's suggestion is that perhaps the human brain is not a Turing machine (i.e. it is not a computer, nor can a computer be like it), and that perhaps there's something quantum going on inside our heads.

If so, then there's no hope for today's computers (for that's all that these machine learning / AI systems are) emulating the human brain. It might be that they mathematically cannot have truly human characteristics like imagination, universal adaptability, etc. It'd take a significant break through in quantum computing (that's wild arsed guess on my part) to begin to get something plausibly intelligent.

[Ian W]

@msbbarratt (https://www.pprune.org/members/241542-msbbarratt)

I can assure you that there are no attempts to reduce the safety standards or certification requirements for autonomous or remotely piloted aircraft; indeed it is quite the contrary. Considerable efforts are being expended to research the different system and safety aspects of unpiloted aircraft in the various xAAs and standards and safety bodies. This includes generation of appropriate additions to certification testing and processes. Some of these efforts are (and will be) translating into better safety for piloted aircraft.

Current FMC/FMS should not be confused with those that will be needed for autonomous flight. In manned aircraft FMC/FMS designers can avoid attempting to cope with the more difficult emergencies and just hand control back to the pilot. The designers of autonomous systems have no such leeway. Some designs from military systems that cope with battle damage and similar will no doubt port to the civil implementations.

There is considerable industry pressure to implement UAS as passenger carrying autonomous aircraft. Yes, some of these ideas are being pushed by people with little grasp of the problems - people that are only now discovering saying not above 400ft is actually not a useful spec as the immediate question is from what datum, do not understand the complexities of some basic flying and ATM operations. At the same time there are those that do understand these issues and are convinced that the delays are due to bureaucracy and regulation rather than technical.

As an example:
https://www.uber.com/elevate.pdf

https://www.uber.com/info/elevate/

I expect that in 20 year's time there will be autonomous aircraft integrated into the normal airspace. There already are such aircraft in many respects as a UA with a lost link is an autonomous aircraft.

[tdracer]

As is often the case with new, ground breaking technology, it'll be the military that leads the way. There are obvious benefits to getting rid of the pilot on military aircraft - not only do you not put the pilot in harms way, no pilot means lower cost/lighter weight and increased maneuverability since you don't need to worry about the g-loads on the pilot. Heck, with that fancy helmet on the F-35 that lets you look through the aircraft, all it would need today is a secure, high speed data link. But you don't want your $100 million aircraft to crash if you lose the data link, so there would be simple, return to base routine. But eventually that simple, return to base routine would get increased capabilities, eventually getting strike/dog fighting capability and finally fully autonomous. Eventually, when unmanned military aircraft have a better safety record than the piloted ones, there will be the final push to extend it to commercial aircraft. It won't happen soon, but I have little doubt it will eventually happen.
As Ian notes, we're not talking today's FMC and auto throttle/auto thrust type systems - for the most part they are created assuming human intervention. A better example would be FADEC or FBW - systems that can't readily be over-ridden by humans.

[Flying Binghi]

...Vanity

Vanity is a dangerous thing in this business, and it's present in the self driving car industry. To illustrate the "problem" in the self driving car industry, consider the possibility of autonomous cars being "bullied" by humans (they won't drive into me, so I can intimidate it!). When asked on BBC Radio recently (Tech Tent, 10th Nov 2017), an industry personality was deadly serious about solving this problem with laws. Seriously? It becomes illegal for you to act in a way that is interpreted as a danger by someone else's lame brained self driving car? No way! How vain is that, expecting everyone else to be compelled by law to account for the nature of one's own product!


"Vanity". Seems more like arrogance.

No need for more laws as there are already numerous autonomous 'vehicles' operating that can be "bullied" by humans and methods available to stop the bully's. A prime example is a lift in a building. Yer summon it and tell it where to go though the lift operates the door and moves under its own computer control. If somebody comes up to an open lift door and holds their arm out they have effectivly stopped the lift in its tracks. Normal polite practice if yer holding a lift for some late arrivals and rarely bothers the people already in the lift. The "bullied" comparison would be when some drunk or yobbo comes up and keeps blocking the door. Comparing it to autonomous cars being bullied I'd suspect the police would find more then one matter to remove/control the bully's

[msbbarratt]

@Ian W,

Yes I'm aware that the established traditional practioners in the field of flight controls are working very hard indeed to do things the right way, to their immense credit. And they do it the traditional way; carefully evolving a testable specification, applying analytical forethought and rigour along the way, and then testing it.

That is how it should be.

It is indeed the newer breed of transport entrepreneur who I worry about the most. They seem to see rules as barriers to be pushed over, rather than enablers of a far larger market.

Unfortunately I can't guarantee that politicians won't be persuaded to make rash decisions. So far in the self driving car endeavour politicians have behaved quite sensibly (California publishes the statistics resulting from trials, much to Google's annoyance...).

Uber's approach to developing self-driving technology is a wonder to behold; it seemingly involves zero forethought and specification writing, being nothing more than throwing a lot of data into a machine learning algorithm and accepting whatever emerges as "finished" so long as it seems to behave itself!

I can add another example: https://www.theregister.co.uk/2017/05/17/unmanned_traffic_management_drones_uas_rpas/. Mike Gadd of the CAA did a good job of pouring a large quantity of cold water on their parade by reminding them that their software would have to be considered "safety critical", and therefore certified as such. It rather took the shine of their glitzy animations and exciting talks...

[Pugilistic Animus]

It sounds just like a HindenTanic...new ideas are dangerous and many new ideas are really old ideas..

[Ian W]

I think that Mike Gadd was saying that if the UAS Traffic Management System (UTM) 'took over control' of a UAS then it was acting as pilot in command and the actions could be considered safety critical.

One has to split out the vehicle software from the UTM software. Very much as in the current ATC system the aircraft may be under mandatory control, but the pilot remains responsible for the safety of the aircraft. The same applies with the autonomous aircraft - the certified autonomous software is responsible for the safety of the aircraft and not the uncertified 'safety related' but not safety critical ATC systems.

The UTM systems are doomed to fail as they are not 'integrating' UAS into the system they are setting up UAS reservations. They are doing this without any regard to the other users of the lower airspace, the powered parachutes, microlights, hang-gliders, crop sprayers, helicopters etc. etc. The lack of regard is due to abject ignorance of the operations in the lower airspace by people who think that throwing money solves problems. These UTM systems though should not be confused with integration of full size (i.e. greater than 55 pounds up to Heavy) autonomous aircraft into the airspace managed by normal ATC systems from low level up to above FL660.

[msbbarratt]

Absolutely.

The phrase “extended flight controls” by Mike Gadd. was used. I take that to mean that the combination of whatever subsystems are on the aircraft being instructed by a ground based traffic management subsystem results in a single distributed flight control system, split between ground and air acting as one.

If so that would hint at a perception that an automated traffic machine management system issuing instructions to an unmanned aircraft in effect becomes part of that aircraft.

That would also mirror ATC. Yes, the pilot of an airliner has the ultimate responsibility. But the ATC officer on the other end of the radio and the system he/she is using both have to be certified as fit for purpose.

Take the pilot away and the ATC officer and their systems just became a lot more important. Replace the ATC officer altogether and it’s got even more “safety critical”.

[alosaurus]

The Head of Airbus Engineering Bernard Zeigler said this was an ambition of Airbus in the early 80's. The A320 was the first step...back then the design concept of the A330/340 was intended to be single crew. If the technology existed today it is 12 years before an aircraft goes from the board to Airline Service. Not in my working life.

[Ian W]

The 747 went from forming the design team in 1963 to commercial service in 1970 and millionth passenger 6 months later. 747 Timeline | Boeing 747 (http://www.boeing-747.com/747_timeline.php)

Perhaps the 747 team didn't know it wasn't possible as they did that with slide rules and manual technical drawing before CATIA CAD software was available to 'speed things up'.

[riff_raff]

It is indeed the newer breed of transport entrepreneur who I worry about the most. They seem to see rules as barriers to be pushed over, rather than enablers of a far larger market. Excellent point. Most of these "flying taxi" projects I've seen don't seem to appreciate the reality of how difficult it will be getting their vehicle design certified for passenger service in most countries. They seem to think entrenched government regulatory agencies like the US FAA will readily revise established procedures and regulations just to accommodate their business plans. Even getting type certification for a conventional new aircraft design under existing regulations can easily take 3-4 years.

If you take a look at the Uber Elevate white paper, they plan to start with a piloted 4 passenger vehicle.

[ShotOne]

And it's not just the regulation but practicalities too; I'm a strong advocate of drone technology where it's appropriate. I have one myself. It has guards on its rotors...one of which is now broken! Even on trains, where speed is the only operator controlled parameter, the unmanned option is not generally accepted by the travelling public.

[PEI_3721]

A pilots view.
Whilst I don’t agree with all of the points, the overall theme is very useful.
https://airlinesafety.blog/2016/01/31/are-pilots-going-to-be-eliminated/

A theoretical view of some of the problems - myths - traditional stories - a widely held but false belief or idea.
http://cmapsinternal.ihmc.us/rid=1M06PB7MQ-Y4BTVG-1C1V/51.%20Myths%20of%20Autonomy.pdf

https://www.linkedin.com/pulse/automation-intelligence-autonomous-machines-human-factor-moin-rahman

And a military view:-
“Overall, the Task Force found that unmanned systems are making a significant, positive impact on DoD objectives worldwide. However, the true value of these systems is not to provide a direct human replacement, but rather to extend and complement human capability by providing potentially unlimited persistent capabilities, reducing human exposure to life threatening tasks, and with proper design, reducing the high cognitive load currently placed on operators/supervisors.”
https://fas.org/irp/agency/dod/dsb/autonomy.pdf

Blame is so satisfying; without a pilot who do you blame.
The public might only accept fully autonomous commercial aircraft when they can identify someone, at the sharp end, to blame.

[Fire and brimstone]

Pilotless future??

"Bring it on - can't come soon enough".

Even cheaper air tickets, and no more human rights abuse of pilots.

Win-win.

[clareprop]

It will come - inevitably. Just as it has for lamplighters, typists, assembly workers, stock traders, hotel workers etc. 'Hold on', I hear you cry, 'they're not exactly skilled employees!' OK, so how about the computer operators, programmers and analysts employed by big companies when Burroughs/Univac/NCR/CDC/Honeywell and IBM ruled business information just 30 years ago? What about the engineers who used to actually visit customers when their mainframes broke?
'Not really the same as pilots though'. So what about the train drivers looking forward to a career on the DLR...drone pilots in the military? So many examples if you look.
It won't happen for commercial pilots on here now - or probably the ones looking to start a career in the next five years...but happen it will.

[tdracer]

Even on trains, where speed is the only operator controlled parameter, the unmanned option is not generally accepted by the travelling public. Yesterday, about 70 miles south of here, on the inaugural run of a new "high speed" train between Seattle and Portland, the train derailed. At least 3 people dead, 80 injured, and the main southbound freeway between Seattle and Oregon is still closed.
The human controlled train entered a 30 mph corner at 80+ mph.
Everyone is asking why the (currently available and certified) COMPUTER system to prevent such a human error wasn't installed...

[Tango and Cash]

Self-flying airplanes are closer to reality than self-driving cars. Not that either are close to reality IMHO. Commercial airplanes operate largely in a controlled environment, under control from ATC, and with other commercial airplanes which follow the 'rules of the sky'. Yes I know there are lots of exceptions and emergencies/failures (largely solved without drama by experienced, professional human crews). But for the most part, the in-flight behavior of commercial airplanes is fairly predictable.

Self-driving cars have to deal with an environment several orders of magnitude more crowded, full of human users who are 1. not communicating, and 2. not following the 'rules of the road'. Most of the drivers around here seem intent on violating as many driving regulations as possible in the most unpredictable way possible. How is any programmed system (or even a learning AI system) supposed to figure out what some muppet intent on his next text will do in his ignorant bliss?

[Boeingchap]

A large factor of pilotless aircraft is the cost of the aircraft and the reliability of the systems in the aircraft - the cost plummets and the reliability increases
All of the current design and systems are done because there is a pilot , the control loops all have to go to a cockpit , all information has to go to the cockpit and be presented to a human - this adds much complexity and cost ,complexity reduces reliability.....
Think...
there is no cockpit , there are no instruments , there are no throttles , no windscreen, no W/S heating no seat ,no circuit breaker panels ,no dv windows, escape ropes , no flight directors, autopilots , no control column , no flap levers,gear levers trim wheels , no crew oxygen , ipads , manuals , life jackets , no cockpit heating or ventilation , no radios , no radar displays , no wire looms to the cockpit , no armoured cockpit door , no .......you get the picture
systems simplify because you remove the human , he/she is not there to push buttons or look at lights or make choices and actions right or wrong based on information given them by the aircraft.
It becomes a design delight and cheap as chips to do , the aircraft never gets fatigued and can operate 24/7 to exactly the same standard .
now tech develops along different lines , maybe laser guidance , maybe new trim systems or flying controls that automation can control but humans cant , maybe ground taxi systems that avoid collisions ...who knows , but it will be cheaper and ultimately safer .
no hotels,no sims ,no licences,no crew room, no crew transport ,no sickies, no rostering arguments ,no one driving to work , just 1 bloke in ops -lol

[megan]

And no pax.

[Dan Winterland]

I was at a RAeS event recently where this was being discussed. One presenter, an engineer was enthusiastically proposing that pilotless airliners would be here in 20 years. I disagreed, and when asked why, I pointed out that as a pilot, I consider that only about 5% of my flights proceed as planned without requiring some form of intervention. I fly in a high threat environment with long periods of poor weather, high terrain, less than sparkling ATC, political issues and a high number of passenger handling problems. Nearly all my flights require me to apply my considerable experience and presence on the flight deck to resolve.

And not forgetting that the EICAS/ECAM procedures are an engineers best guess at the problem. In my experience, rarely do they work as planned. In the last four big events I have had in the last five years on Airbus types, none of the ECAM procedures correctly dealt with the actual problem. The last one would have had us depressurising the aircraft had we followed the EACM procedures. The solution came from our technical knowledge of the aircraft and analysis of the actual problem.

Of course, these issues could be resolved with more computing power and more procedures. But if this were to happen, any pilotless airliner will have to have to be equipped with artificial intelligence which would have to apply a variable risk assessment factor to resolve any problems. This may even have to ignore inputs from ground control sources if 'thought' inappropriate. I reckon it's going to be a long time before that level of AI becomes available, and even longer before manufacturers, regulators and insurers are willing to endorse that. Let alone the fare paying public accepting it!

One question asked was "Would you get in a self flying airliner?" My answer was that I wouldn't have to as I consider that they are at least 40 years away and my passenger flying days would be over by then, if still alive.

[Boeingchap]

EICAS/ECAM -would not even be fitted to an automated aircraft, -at present the quantity of information presented is restricted , because a human has to interpret it and procedures are there because the human misinterprets the information given.
It not so much as an AI system ,as a pure logic system , a bit like a chess computer -think of 1000 engine parameters being used ,could the pilot analyse them and then what could the pilot do ? his actions are limited -automated responses can be much more complex ,engine control functions much more complex, way beyond simply moving the throttle or shutting off fuel -too much for a human to calculate
The tech would move into all the control systems on the aircraft ,each improving massively
The question will become who is dumb enough to rely on a human up front , you cant even tell your altitude ,without the machine telling you it , let alone maintaining it within 10 foot, without instruments or an autopilot .....it will all happen sooner than most think

[wiggy]

Ah, that answer to it all, “A.I.”

Talking about this thread to one of my offspring who has done just a bit of work and research involving “AI” (ground road traffic environment, traffic signalling, traffic flow decision making, that sort of thing ) her view was that the capabilities of AI are frequently overstated. It’s improving, sure, but still “clunky” ....but neverthless sold as the answer to everything by the enthutiasts.

In her VHO truely pilotless/ AI driven routine commercial flight is still multiple decades away.

[RVR800]

Passengers PROBABLY would not travel in an bus without a driver. They DEFINATELY wouldn't travel in an Air-bus at 6 miles above the ground without a pilot. Buses will stop so that the driver can rest or when they are ill. Stopping is simple. A co-pilot facilitates longer flights. Airline automation is a long term plan to come many years after buses are automated ...

[tdracer]

They DEFINATELY wouldn't travel in an Air-bus at 6 miles above the ground without a pilot. I beg to differ. There is considerable evidence that a significant portion of the traveling public would do so in a heartbeat if it was $20 cheaper (or 20 pounds - sorry but I don't know how to create that little symbol on my keyboard)...


Wiggy, I agree with your offspring. AI in it's current iteration sucks (I have a few stories regarding AI on cars) - but is getting better pretty much by the day. I'd say we're between 30 and 50 years from truly autonomous passenger aircraft. It won't happen soon, probably not in my lifetime. But I have no doubt it will eventually happen.

[dual land]

I wonder what portion of today's traveling public has jobs when AI can deal with air traffic's level of complexity. This of course isn't the only industry affected by it. Maybe they really will need a 20$ reduction in fares in order to afford them.

[Highflyer3]

Here are a few of my top reasons we will not see this within the next 30-60 years:
1)Today pilots mainly monitor the aircraft computers, what will be the point in having a computer monitoring a computer?
2)Passengers will not go near the aircraft (most) and the first airline to buy it will most likely go into losses scaring off other airlines.
3)The change in inferstructure would be huge, we would have to change all the laws, the way airports work and many airways etc!
4)The technology we have today cannot create general AI which is what we need as it will think like a human. (And won’t exist for a few decades)
5)A big one is testing, this new aircraft would have to go through years, even decades of of testing and it would have to fly almost errorless to be accepted.
6)Pilot unions will put up a BIG fight against it delaying it happening.

There are many more but also the media is mainly trying to make a story to make money, Boeing basically said they’re working on the building blocks fo this technology if its needed and they simply brought a drone company (along with airbus.)
Even IF this happens in our lifetime anyone reading this will be an experienced captain so the remains jobs will easily be take by us. (Most likely anyway.)

:)

[Ian W]

Air traffic is an entirely different matter to autonomous UAS (aka drones). The current problem with Air Traffic Management systems is that they are based on a 1930's air traffic control concept procedural control. All attempts to update the concept are mired by 'experts' in the 1930's concept who can only think in its terms. So trajectory based operations where an aircraft's future 4D positions are known allow the aircraft to fly in a way that is considered efficient by the operator (what Europe is calling the 'business trajectory). This is simple for automation to deal with and reduces airspace congestion and has been demonstrated in large scale research. However, the implementing ANSPs do not understand the concept and try to force 'trajectory based control' into their 1930's procedural concept and it becomes 'time based control' (the 4th dimension) and unsurprisingly the problems seen in today's system remain. There's more but it's not for this thread nor this log :)

[Bergerie1]

Ian W,

You are so right! Not only would 4D trajectory management increase airspace capacity it would also be far more efficient for all airspace users. But, as you say, the ANSPs do not understand the concept and are stuck in a 1930s mindset.

[Ian W]

Highflyer3

Taking your points in turn:

1. Current automation was designed a minimum of a decade ago based on analyses a decade before that - development is slow. Think how much your brick phone has advanced to a 'smart phone' in that time. The systems analysis was that it was easier for the programmer to drop out of problem solving early and hand the bag of bolts to the pilot - who was only using the FMS as a labor saving device. Anything difficult - automation give up.
Significant advances have been made since then - automation that can fly a very badly damaged aircraft almost as if it were undamaged. AI has increased to a level not thought possible - a computer was given the rules of chess and within hours was beating grandmasters see https://www.theguardian.com/technology/2017/dec/07/alphazero-google-deepmind-ai-beats-champion-program-teaching-itself-to-play-four-hours
There may be triplexed computer systems this is quite normal.

2. I would suspect that Company caution would see the freight dogs being automated first possibly as single pilot systems. I can remember when elevators always had an operator. I assure you just like automated trams, trains and cars; automated aircraft will not worry future generations.

3. The autonomous aircraft should appear to everyone else to behave and operate like a manned aircraft following the same rules. There should be no changes just to accommodate the unmanned/unpiloted aircraft. Any aircraft with a remote pilot must be able to operate safely if the link to the remote pilot goes down - so all should be capable of autonomy.

4. See the chess playing robot story above ( https://www.theguardian.com/technology/2017/dec/07/alphazero-google-deepmind-ai-beats-champion-program-teaching-itself-to-play-four-hours ) AI has changed a lot in the last 18 months; its rate of change and advance is faster than you would believe.

5. Certification will be an issue particularly where the certification tests were set up for human flown systems. This is going to be a complicated area not because of the length of time for testing individual aircraft but the time required for international (RTCA/EASA) agreement on what those tests should be. This work is already starting.

6. True as some did for glass cockpits and look where it got them. The reason I started this thread was to get the pilots here thinking about what WILL happen in the future. Just saying "No! not in my lifetime!" is not an option. There will be a requirement for pilots for decades to come - but do not think that UAS will not be sharing the airspace with you - they already are.

[Derfred]

Ian W,

Google's chess playing system AlphaZero simply used "machine learning". It was told the rules of chess and then played against itself a gazillion times, randomly initially, with a simple reinforcement algorithm by which it eventually "learned" good moves vs bad.

That's just statistical analysis of big data. Something computers are good at. Artificial Intelligence it ain't. And it certainly takes us no further towards the holy grail of Artificial General Intelligence.

To translate that to autonomous aircraft would involve crashing a gazillion aircraft in real world conditions until it "learns" how to not crash. Sorry, but playing chess isn't transferable to a dynamic safety system in the real world.

In the global race towards autonomous cars, not even the companies themselves are in anyway pretending their vehicles will be "intelligent". They won't be. They will be following simple algorithms based on big data. When they get into trouble, they will just stop, much to the annoyance of human drivers (I'm sure we'll hear a lot about that in the coming decades).

Aircraft can't stop.

And to put any timeframe into perspective, I still have to read 50 pages of NOTAMs everyday that are still formatted for teletypewriters.

[Highflyer3]

Lol I remember playing droughts with my iPod in 2008 against AI and loosing... hasn’t changed than much. Additionally I’m not sure what AI will do after a total electric failure, or when little kimmys team of hackers in North Korea hack them. So far in my life I haven’t seen a driverless car (a thing that goes forwards backwards left and right) or a driverless train (something that goes forwards and backwards.) Planes may becomre automated in 30 to 50 years but it will happen slowly a pilots will disappear SLOWLY. Also if you listen to the Boeing VP interview he makes it clear that it’s an option in case we need it for commercial aviation. Additionally airbus said they’re looking at making autonomy for sign pilot operation, they have said they’re looking at how to develop AI but haven’t directly said they’re working on a pilotless future (It’s just the stupid media trying to earn money.) I also havet seen a single concept or future design without pilots. I’m sure we will see Unmanned cargo and fuel planes, and urban air taxis etc but it’ll be quite a long time untill pilots are non existent. Like ADFUS said by then humans will have lost most jobs anyway by then.

(And also where did you hear EASA working in the testing requirements for autonomous aircraft?)

[RudderTrimZero]

In the world of management, thousands of knuckle-heads get paid for doing nothing. In the world of flying, due to automation, thousands of pilots get paid for doing nothing. We are finally on a level playing field.

[Ian W]

Highflyer3

I am surprised you haven't seen a driverless car or train, you could go to the Docklands Light Railway not far from you and see one that has been running for years. Driverless cars are common place in many cities in the USA and are approved for trials in UK too.

https://learningenglish.voanews.com/a/dubai-plans-to-launch-a-self-flying-air-taxi-by-july/3726018.html

There are plenty of these unpiloted air taxis that are actually past the concept stages and in development. Optionally piloted aircraft have been around for nearly 10 years. Boeing: Unmanned Little Bird H-6U (http://www.boeing.com/defense/unmanned-little-bird-h-6u/)

All you need to do is look at the work going on in EASA and RTCA on standards for UAS/RPAS/Drones all of them by definition need to provide certification standards for autonomy as the UAS/RPAS is autonomous when it has a comms link failure. All is available on the internet with a simple search.

The arguments being made here are that pilots are there to solve all these issues that automation cannot solve. Yet we also see discussions here on deskilling by letting the automatics fly the aircraft so when LOC happens the pilots are unable to regain control. The software is then written next time to not hand that LOC to the crew but to solve it instead. When the automatics become better than a deskilled pilot in recovering non-nominal events perhaps even identifying ahead of time when they will happen and mitigating them in advance, then it becomes very difficult to justify having a pilot.
The move from pilot in the loop to pilot on the loop has already started. once all aircraft systems and operations are pilot on the loop, the next step is pilot out of the loop. (and pilot is already out of the loop on things like FADECs)

There is a defense contract being competed for in the US for carrier borne UAS tankers that will autonomously operate from carriers and provide range extending refuelling to USN and USMC aircraft in combat operations. Refuelling and carrier landings and takeoffs are more challenging than anything that airliners or freighters will experience.

[wiggy]

Refuelling and carrier landings and takeoffs are more challenging than anything that airliners or freighters will experience.

Really? At one level they all seem (mechanically) to be fairly straightforward 3D processes, though perhaps for any human element involved needing nerves of steel at some point in proceedings.....as for refueling ( I assume you are talking air to air) most navs I flew with were convinced that AAR (hose and drogue) involved just six "commands" i.e. up/down, left/right/ forward and " :ooh:..

I'll accept we are getting close to genuine autonomous airline ops being just around the corner when we've seen routine automated operations with 737/A320 sized aircraft between basic airfields with mostly visual approaches (such as those found in the Greek islands, elsewhere in Europe and probably all over the States).

[Highflyer3]

Ian W,
I see what you mean now. When I fly my drone Andy loose signal my drone will either autonomously fly back to me, hover or make an automatic decent in emergency conditions. This is what I think you are talking about. My drone is not large but this can relate for other larger drones. If my battery goes bellow 10% the drone will fly back And allow me to take control when it has calculated the minimum distance needed to land and if it’s no longer ctritical. This is very good and all but this is fairly simple autopilot. It follows its gps to the takeoff point using altitude from where it took off and will fly above my minimum set altitude however it will not avoid obstacles (so,e drones will do that.) I can also turn on the auto land feature if I wish however for more hard landings e.g close proximity to obstacles I have to land as it is not intelligent. This will work on small recreational drones and air taxis however all air taxis have parachutes and are like a slightly more developed autopilot following Pre established routes as they are short distances. This is different to large jet commercial flying. It’s like having a jet ski that follows a route via gps and avoids shallow water and rocks using sonar and cameras. Compare that to a large cruise ship where it’s a very different type of handling etc. Autopilot is different to AI. Autopilot is programmed by pilots and has a minor level of intelligence to sustain flight. When there is a storm unless told to do so it will not avoid it. In order to just make th AI to fly an airliner it will have to cover EVERY possible scenario and know all the runways and approaches etc. I’m not sure as to how they’re going to make this by making AI programmed to fly planes or just by having deep AI to act like a pilot would. Either way I dislike AI for many reasons (other than flying.) But that’s a bit of topic.

I read the EASA regulations and they’re on about aircraft not with a pilot on board, but a controller on the ground and does not mention air fart with no human interaction at all, they’re supporting drones. Not as the media calls “robot piloted planes.”

Also another point I forgot to add is Eurymenko got fired as Airbus believed he was pushing innovation to single pilot operations/autonomy to fast. It shows how Airbus don’t believe we have the technology to do it yet as well as the time it will take to do this will be longer than he proposed. As well as Boeing also said they’re working on the “building blocks” of this technology INCASE WE NEED IT. It’s still not even certain if we will have, although most likely at some point in this century and certainly in the next.

[Ian W]

wiggy

The UAS will also be playing chick as well and buddy-buddy refueling can be interesting.

With GLS there is really no need for any 'visual'/non-instrument approaches to any runway in the world. Add the SVS that automatic systems can use as cross match to 3D radar maps and automated approaches should be possible to anywhere. With GLS an aircraft can be 'established' while in a turn on a GLS approach so the procedures can be a little more interesting than just long straight in descents.

[KayPam]

SVS/EVS is also very difficult to put in place in piloted aircraft, let alone unpiloted ones...

[His dudeness]

2. Without 1., (above), the international insurance market won't touch it, no insurance = no fly.

There is not a single insured nuclear power station, yet 449ish of them are working around the globe...

[Marchettiman]

So, can we see this scenario happening?

Autonomous Boeing and Airbus aircraft prototypes begin an extended flight test programme, airport and ATC infrastructures do the same, say 10 years for proof of concept and then another 5 years before ICAO and national CAA's can agree rule making for the new technology.

In the meantime word gets out that pilots are going to become an extinct species, save in the poorly paid General Aviation sector so the attraction of a career as a commercial pilot begins to wain. Banks (including that of mum and dad) no longer see an investment of £100k+ in training a worthwhile proposition so the supply of new pilots begins to dry up and experienced ones opt to cash in on their earliest retirement dates.

The pilot shortage situation becomes a desperate one, salaries increase and the low cost airline economic model becomes redundant which forces up fares and pax numbers fall, perhaps drastically. Big name operators gobble up their balance sheet reserves before reverting to the legacy airline model. Traditionally piloted aircraft lose value like diesel cars, and the banks who have financed them shy away from the airline sector with extremely burnt fingers.

Investors in businesses get scared by major change, but of course there are always investors for new ventures that embrace upcoming technologies. What happens in the transition period?

[parabellum]

There is not a single insured nuclear power station, yet 449ish of them are working around the globe... Meaning the owning authority have chosen to self insure and they only have to answer to themselves as, hopefully, their power station is not going anywhere that could involve third parties who might insist on a very high level of third party cover.

It won't be the hull insurance cover that is the issue with a pilotless aircraft, it will be the passenger and particularly third party liabilities cover that other countries/airports etc. will insist upon before over flight or landing rights are granted. Not sure about today but the benchmark for assessing the level of cover required was two jumbos in a mid-air collision over the CBD of a major city, it runs to billions of dollars.

[ex-Dispatcher]

As I've stated before when I make very occasional posts, I am not a pilot and will not comment on piloting issues - I am a Position Navigation and Timing Technologist with particular interest in testing navigation systems with emphasis on security (and of course evaluating the effectiveness of highly automated systems and how to test them). I suggest that some of you might appreciate reading the NTSB investigation into the recent (and fatal) accident between a highly automated car and a tractor/trailer in Florida. It highlights many of the issues associated with autonomous vehicle navigation - and also the issues around developing a navigation system suitable for use in dynamic traffic situations that does not require human intervention at some point.... As I have less than 10 posts I can't provide a link to the report but its reference is NTSB/HAR-17/02. If you think it worthwhile, perhaps someone else could post the link?

[msbbarratt]

You're quite right to cite that report (https://www.ntsb.gov/investigations/AccidentReports/Reports/HAR1702.pdf <pdf>). Keep posting - you'll soon get up to 10.

One thing is for sure. If we do end up with pilot-less airliners, we'll get an accurate count of how often advanced high tech automated avionics systems fail; just count the crashes.

[tdracer]

I suggest that some of you might appreciate reading the NTSB investigation into the recent (and fatal) accident between a highly automated car and a tractor/trailer in Florida.Somewhat apples and oranges - the Tesla system was not designed or intended to be fully autonomous - it was a "Level 2" 'supervisory' system intended to assist the driver, not take his place. That he used it as a fully autonomous system so he could watch a Harry Potter film earned him a Darwin award.
None of the automotive systems currently for sale are capable of fully autonomous operation (aka "Level 5"). The best systems currently for sale are Level 1 or 2 - with some Level 3 currently in Beta test.


Edited to add this link that defines automotive automation:
https://www.caranddriver.com/features/path-to-autonomy-self-driving-car-levels-0-to-5-explained-feature

[Derfred]

With GLS there is really no need for any 'visual'/non-instrument approaches to any runway in the world. Add the SVS that automatic systems can use as cross match to 3D radar maps and automated approaches should be possible to anywhere. With GLS an aircraft can be 'established' while in a turn on a GLS approach so the procedures can be a little more interesting than just long straight in descents.

Ahh, yes Ian W... the point is not how well an aircraft can autoland at a GLS-equipped airport with everything functioning correctly. Or what the USN are doing with UAS's. No one cares if one of those ditches in the ocean (except the USN beancounters).

The point is how well an aircraft can autoland at an airport where the GLS/ILS just failed, or when it has to divert to an airport without one, with degraded systems, with passengers on board. And how well it can make the correct decision to do so.

Talking about autoland is a bit of a waste of time, we've been doing that for many decades.

Just taxiing to and from the runway autonomously would be an infrastructure nightmare to develop, let alone any airborne decision making about weather and fuel.