"Pilotless airliners safer" - London Times article

[RVF750]

Have a think of this...

Right now, we can hear of a crash involving aircraft "A" or " B" and turn up at the airport next day looking at lines of "A" and "B" on the ramp and still willingly get in and fly (in) them.

Well, the crash was the pilots fault, and these will have different pilots so that's ok, isn't it? worst case, they'll look at the reasons for the crash and try not to make the same mistake themselves.


Step forward into the future. Now, when the airline rep says step up and fly, who will there be to blame for the crash yesterday? Are today's planes flying on Version 3.1 or 3.1.1? And what version was on the one that crashed? Until they can get round this issue, only the deluded fools who designed the system will be brave enough to fly on it.

Also, Google car. It's managed 170,000 miles of safe driving, but only on the roads around their headquarters that are in it's very complex mapping system. Put a set of temporary traffic lights up and it grinds to a halt. The AI is so far off I'll dead and gone before it ever comes down my high street.

[Tourist]

Aluminium

I suspect that you are referring to me with your comment about pontificating?

If so, what makes you think you know my background?
Does Airline, Corporate, Military rotary and Military fixed wing qualify me to pontificate?
Does having operated with and in extremely close proximity to RPAVs for many many hours over Iraq, Afghanistan qualify me to have an opinion?
If not, then tell me what your qualifications are?



Those who believe the technological challenges are insurmountable should ponder on this.

Various militaries in the world are testing autonomous UCAVs
These are unmanned, autonomous combat aircraft.

Autonomous X-47B Flies In Formation With Fighter Aircraft | Popular Science

These aircraft must find must take-off (from a carrier in this case) navigate, find and kill the enemy, navigate home and then land on.
No beacons, no radio control, no atc, no TCAS.

Many believe that the F22, F35, Typhoon etc will be the last generation of manned fighters.

Few on here would pretend that an airline pilot needs to be or is as highly trained or flexible as a combat pilot in the military. Militaries all over the world spend years and millions of pounds selecting and training each military pilot.

Yet for some reason the militaries truly believe that a computer can do the job better. Even though an average military mission is vastly more complex and changeable than a civvy flight, with vastly more variables to deal with and vastly more tasks to accomplish such as air to air refuelling, deck landings, combat, combat damage, defending, attacking, ECM, ESM, formation flying etc etc
For some reason those idiots in the military truly believe that they can do all that without a pilot.

What fools they must be....

Unless you truly believe that the job of an airline pilot is somehow a greater challenge.....?

As with many things in technology, things trickle down from the military, and compared to what they are trying to do with combat aircraft, airliners are childs play.

[Tourist]

Ozy

You say that they have not even got a detect and avoid system yet.

Remind me what a current airliner uses at the moment?

Oh, that's right, eyeballs and TCAS, or just TCAS in IMC.

Well TCAS can be fitted easily to a UAV and it won't make a mess of it like humans, and both visual and radars are normal fit to even older generations of military aircraft so no problem there.
UCAVs are having systems designed to enable air warfare against fast moving stealthy opponents.
Avoidance of transponder equipped aircraft is again mere child's play in comparison.


Turkish

Your use of the term "AI" shows your lack of understanding of what is trying to be achieved.
Nobody is going to give them a Turing test, any more than you test your FADEC or Fly-by-wire.

Plus google autonomous cars. Since you brought them up I feel it is only fair to point out that they are doing well...
There are many on public roads in Florida California Sweden Japan and in January guess where? The UK!
http://www.economist.com/news/specia...-look-no-hands
http://spectrum.ieee.org/transportat...lfdriving-test
http://www.theverge.com/2014/10/20/7...-on-the-planet

[Aluminium shuffler]

Tourist, the UAVs are currently able to fly a pre-programmed route and be given updates while in flight, but can't independently replan a route. Nor can they do any kind of complex air combat manouver, though I would cede the point that if the airframe is strong enough, it could out-turn a human pilot in a simple manouver to escape a guns kill. That ability will eventually come, but as I clearly said, current technology is not intelligent or flexible enough to cope with simple changes in circumstance, and outright manouverability is not much use to an airliner. What is needed in an airliner is a command system that can operate on faulty or partial data and still get the job done. Like I said, I had an autoland the other night that was going to put down on the grass, and manual intervention prevented that. That won't show up in any collated data of automation vs manual skills because there is no forum to collect such statistics as no accident or incident occurred, and that is the point Ozzy and several others have made. If you find that automation is more reliable than two human pilots, you're flying with the wrong people!

[ShyTorque]

Tourist, firstly I think you are trolling.

You are concentrating on the technically achievable but ignoring the practical integration of an unmanned aircraft into the existing system. Flying a combat aircraft in a war zone is a totally different environment to a civilian airliner full of passengers operating in proximity to other aircraft, most of which will be under manual control.

How would your fully autonomous helicopter deal with spurious but false warnings such as a tail rotor chip caption (or even double engine chip indication coupled with a main gearbox chip caption, both of which I've seen) over totally inhospitable terrain? I've flown helicopters for over 35 years. Many, if not most, of the "serious failures" I've experienced have actually been spurious warnings.

On the other hand, I've seen inflight total loss of output from both channels of the aircraft's data acquisition system which are supposedly mutually redundant. I've also seen real warning captions that were not in the pilot's flight manual or maintenance manuals. The aircraft manufacturer had to take further specialist advice on what these captions meant; this took some days. How would an autonomous aircraft deal with these situations? The problem is, just like the programming of an aircraft simulator, rubbish in = rubbish out and if there's an off-model event, the outcome of an automatic system is as uncertain as that of any human pilot.

[Tourist]

Aluminium

Try actually clicking on the links I put up.

That Blackhawk is independently is planning a route on video!

Re the bad autoland, what on earth has 60s technology in a system designed to have pilots got to do with this discussion?
I'm not saying that current airliners are autonomous. They are not designed to be.

Shy

The spurious warnings issue is one of the areas where an autonomous airliner has the advantage.

In a modern airliner, the engines are sending vast amounts of info back to Rolls or whoever that isn't even going to the pilots. You can be met on the ground by an engineer that tells you that your engine is tits. The computer can continuously monitor for trends and build up a picture and has a far greater chance of working out whether a caption is spurious or not.

You talk as if integration is an unachievable dream.
It is being worked on and tested today.
Yes combat is different. It is a more difficult problem in just about every way.
Civil aircraft should not and rarely do get into any sort of proximity to each other compared to military. Just ask any military pilot when he first gets TCAS. The response is always "You want me to move for that!? That's not a confliction! It's miles away!!!"

[ShyTorque]

You talk as if you are the only one with any military experience. Also, you haven't answered the question about total data loss or spurious chip warnings.

[Tourist]

Shy

Read my last post. That is all about spurious warnings, though chip warnings are a rotary thing really.

Re total data loss.
An aircraft that is designed to have a pilot in the system relies on him to be there.
Any system designed to not have one replaces him with something else. You are not just going to remove him and leave it as it is.
An autonomous system must necessarily have greater redundancy. There is nothing black magic about it.

An Airbus currently has a bunch of inputs. If it cannot work them out it throws up its hands and gives it back to the pilot like in the Air France crash.

An autonomous system would add in more inputs of which any number of current technologies exist and would have bowled out the problem instantly as pitot problems and the passengers would never have even known something happened.

Current airliners have very limited and frankly archaic data sources in many cases.

[16024]

Right.
Let's agree for the sake of argument that it's theoretically possible. Just about anything is possible given the time, money, will, bloody mindedness etc.
Let's imagine some gambler of a CEO and some trusting shareholders (this is a fantasy "what if" isn't it?) who says, "Ok, why not?".
Now give us a timescale. We've been banging on about this for at least 20 years. Another 10 to make our minds up and put some money down. Another 10 years to get a prototype out, because every single thing is new, as we have already established that our 60's technology won't do it.
10 years possibly loss-making on remote freight-only rotes (loss-making if you have to construct dedicated terminals or airports).
If we start tomorrow, this takes us to 2044. When is the oil going to run out?
Another 10 years of people getting slowly used to the idea, and we are good to go. What's the lifespan of a new design? 20 years before we expect to get to the ADV, NG, NEO life extension versions? That would take us beyond 2074.
It ain't. Going. To. Happen.

[Tourist]

Ozy

TCAS is exactly an automated system with a human thrown in to add errors!
You are not supposed to do anything other than what the computer tells you.
It is extremely conservative and extremely simple. What possible difficulty could an autopilot have in following it?
The pilot does nothing but act as an autopilot on the computers instructions. badly.

Current airliners have no detect and avoid system beyond TCAS. Why are you so insistant that an automated system even have one? Airline pilots barely look out of the windows from one minute to the next, yet you demand something better from an automated system?
Luckily, because the makers know that they will be held to higher standard, they are being worked on and tested.

16024

If it ain't going to happen, why is so much money being spent on making just that happen.
BAe, Lockheed etc are not idiots and don't like throwing money away. Not their own money at least.

[Tourist]

These guys are real players in technology and aviation
They are not chimps.
ASTRAEA Video
They are not doing it for fun and they are taking it seriously.
62million so far, and this is just the UK

[Huck]

We hardly look out the windows? News to me....

Aaahh.. PPRune.... you never disappoint.

I've been watching this argument for 13 years here, give or take.

I'll say what I always say - I'll start worrying when the freight trains that run by my house are unmanned. They've gone from three crew to two. But not one. And that task is three orders of magnitude easier than operating an aircraft.

[ShyTorque]

Read my last post. That is all about spurious warnings, though chip warnings are a rotary thing really.

Yes, I'm aware of that, having flown both rotary and FW, military and civilian.
I mentioned tail rotor chips specifically because you introduced a rotary wing example to the discussion to further your argument about the available technology. So now you want to disregard it because you couldn't provide an answer to the question I posed about it?

BTW, the military have never so far said that they believe that unmanned aircraft are better than manned. The idea of an unmanned machine is to keep the man safe, away from unnecessary risk due to enemy action.

Of course as an ex military man yourself you will recall that the Defence Minister of 1957, Duncan Sandys, obviously did believe that manned fighters should be replaced with missiles but it didn't really work out that way. It's generally recognised by the RAF that his policy did immense harm to the UK military capability and the associated defence industry, in the long term.

An aspect not yet discussed: If unmanned airliners ever come to be, who is going to be willing to sign off the aircraft to be dispatched, with a known defect? I'll bet a pound to a pinch of ***t that you won't find many engineers willing to sign their life away because once that aircraft gets airborne, it's fully their responsibility, there are no further links in the chain. Many of these extremely complicated aircraft (much more so than now, according to our resident expert, Tourist) will sit on the ground u/s for far longer than today's aircraft.

[AirRabbit]

Originally Posted by Tourist

AirRabbit.

Not entirely sure what point you are trying to make so apologies if I have misunderstood, but re the fear factor in the real world, that is exactly what computers are good at removing. They work the same day in day out.

Incidentally, you are all aware that there are lots of aircraft that currently fly around that cannot fly without the computer working?
ie if the computer fails then it crashes?
They seem to be doing fine. Computers are quite reliable when they have to be.

Hey Tourist – no apology necessary … but I think we’re very likely on opposite sides of this particular point. I know that computers are quite good at doing what they do … and that is they take data and they execute the commands that have been programmed to be taken when the data examined is determined to be of a specifically defined value or value range. Some folks have been very successful in expanding the range of data examination but for all intents and purposes those successes are always limited by the methods used to both gather and submit the relevant data for the computer’s examination. However, it remains true that computers don’t “think” and they don’t “anticipate” … unless, of course, someone preprograms the computer to compare existing data with a history of the kinds of developments that have historically manifested with the same values of the data referenced, and calculate some acceptable level of mathematical probabilities for the development of the same circumstances and then adjust the controls in anticipation of those circumstances developing with the same level of probabilities. Unfortunately, the size of the airplane that would be necessary to house that kind of memory and computing power would likely severely limit the number of passengers that could be accommodated on a typical airliner - and none of this even begins to touch on lightning strikes or EMP kinds of interference scenarios. Just yesterday in Detroit (a major US city) an unknown power outage brought dozens of square miles in the middle and edges of the city to a stand-still for several hours. If a similar problem were to occur while airborne … do we just “write it off” as a Mother Nature hiccup?

And, as for "computer controlled airplanes crashing if the computer quits" … as I understand them, the flight envelope control systems on Airbus aircraft always retain flight control when operating under “normal law,” however, in extreme circumstances, like multiple failures of redundant computers, there is a mechanical back up system for pitch trim and rudder. Instead of this kind of mechanical back-up system, newer Airbus aircraft have an “all-flight-control-back-up system called a “three-axis Backup Control Module (BCM).” Additionally, on the newer Boeing aircraft, the two pilots can completely override the computerized flight-control system to permit the aircraft to be flown beyond its usual flight-control envelope during emergencies.

Regardless of the level of sophistication of any specific computer, its success rate is always going to be limited by the data source, the data sequence, the data accuracy, and the resulting successful submission of that gathered data to the computer. Of course the “computing power” (the speed at which data can be input, processed (compared), and a specifically preprogramed response determined, will be directly dependent on the accuracy of the format of the data to be examined, the number of data sources, the determination of priorities of that submitted data, including those circumstances and potentials that may affect those priorities. Again, whatever the response is to be, that too is limited to what has been provided for by the computer’s structure, and how, and to what, that computer is connected. It is obvious that computer-aided airplane control is here and is arguably successful. But, turning over complete control of those airplanes to a computer or a bank of computers with no human intervention is, in my not-so-humble opinion, still some indeterminable distance in the future – if it occurs at all.

[ConnieLover]

Those who, like Tourist, use the military as an example of technology that is ready to be copied by civilians ignore one crucial fact: in the military a certain percentage of losses is acceptable, which are not acceptable in civilian life.

[Oriana]

Should be re-titled as 'A Bean Counter's Wet Dream"

[Huck]

I'm not so sure.

Lots of folks can explain how it can be done.

Nobody can explain how it will be cheaper.

[Piltdown Man]

So earlier this week NASA, who have a pretty good idea about these things, announced that they will be authorising the launch of the first of a series of rockets to prepare for a manned mission to Mars. If computers were better and more capable they would have been sent. Firing a living thing millions of miles into space is awkward and expensive. Therefore, there has to be an advantage in doing so. I'll suggest that people are being sent because they have a greater chance of dealing with what cannot be foreseen. It's certainly not for the ride as the service levels will be worse than that endured by LoCo SLF - except for the fact they won't be bothered by cabin crew flogging scratch cards. Unless if course NASA will try to offset some of the mission costs...

[Superpilot]

No one from the YES department answered my question:

What would you consider to be a safe automated passenger jet? Is it one that can think as well as human (i.e. have the same level of intelligence) or one that can think a billion times faster given the same (known) problems?

[16024]

With the Mars landing, there's an emotional, and PR aspect to it.
"We are going to Mars". Wow!
"We are sending a machine to Mars". Been done before.
As for the real reason why it won't happen. Yes all the recent posters have great obstacle points which the pro lobby can chip away at, and it seems they have invented half-a-wheel very competently.
The real reason is as I clearly stated a few posts back. Ain't going to happen.
QED