Planes without pilots - NYT

Hacker15e said:
Again, the problem here is the assumption that remote piloting is somehow less expensive than human pilots. After all, the whole purpose of replacing pilots with robots is to save money and increase productivity, right? Unfortunately, it is a seriously incorrect belief. The folks that think this haven't ever actually looked at the real aggregate costs of operating these things beyond line of sight.

The BLOS connection (satellite time) alone for a single day's sortie in the aircraft I'm currently working on costs more than a regional pilot's annual salary. That cost doesn't even touch all the other infrastructure and maintenance costs, and given the technology buying, caring, and feeding that technology is not at all an inexpensive prospect.

Maybe someday RPA technology and the satellite bandwidth on which to run it will be cheaper than a pilot, but it sure as heck isn't anywhere close today.
Click to expand...
The flaw in your thinking is two fold. First, once the initial acquisition costs are absorbed it's no different than any other technological leap. There are plenty of news articles where "experts" have been quoted as saying "no one will ever install (electricity, phone lines, cable tv, cell phone towers, GPS receivers, ADS-B equipment) because it's just too expensive."

Once the tipping point is reached, and it won't be long, you'll see the changeover rapidly. Smartphones are a perfect example. I remember when the RAZR came out and half the population said they'd never spend that kind of coin on a phone. Now, people don't blink at spending $700 on an iPhone.

And second, at this time you are correct about satellite bandwidth but that will also change rapidly. The other thing to consider about the datalinks is that in the UAS world 80% of your bandwidth is eaten up by your video feed(s). A few lipstick cameras and your control link use up much less bandwidth making their subsequent subscription requirements much cheaper.

It would even be possible to piggyback onto an existing ground based network with remote outstations using technology cheaply manufactured and acquired today.

The point is, for those of us who know how it works, we know that we will see the single pilot airliner before my son is eligible for his ATP.
 
z987k said:
Lots of deer jumping out in front of you at FL350? What about snow/(black)ice/gravel? Then there's other drivers, who have almost no training at all, and if we're honest there are virtually no rules in driving compared to flying. Certainly not many people care to consistently follow.

Think about it. Would you rather drive 100 miles in a blizzard or fly through the same storm? We both know which is safer. For me, flying is a lot easier in the worst weather than driving in moderately crappy weather.
Click to expand...

Hit a dear, hit the brakes. Hit a flock of geese, end up I'm the Hudson, swimming because you got extremely lucky.

I'll take the flying, once I got in the air. Because I can go around the weather, not through it. But once your in the weather, I'll take diving any day off the week, and twice on Sunday. Working in three dimension, trying to pick your way through it, and having no choice but to continue through it vs just stopping and waiting.

Flying isn't hard. But driving is a whole bunch easier, even in bumper to bumper traffic on the 405 freeway, through the rain, like my drive home from the airport was last night.
 
esa17 said:
The point is, for those of us who know how it works, we know that we will see the single pilot airliner before my son is eligible for his ATP.
Click to expand...

Read this as, "from my extremely biased opinion, it'll be here in the next 10 years."

No, it won't. Does the tech exist, sure it does. Will the public accept it, will the regulators allow it, and will the programmers be able to write code that satisfies everyone? Most likely not. It's more, much much more than just the capability to do it.
 
ATN_Pilot said:
Yes it is. Not even close.
Click to expand...
Which is why we developed widespread UAV use over a decade ago and self driving or even remotely controlled cars still aren't really a thing?
Sorry, but it's massively easier to program something to fly within the confines of the NAS than it is within the road system where there are probably 1,000 more variables. I mean the sensor array to figure out what is going on in the road system is going to be insane. You need almost no sensors to make a UAV.
From a programming and computer logic standpoint, flying in the IFR system is stupid easy in comparison to driving.
 
z987k said:
Which is why we developed widespread UAV use over a decade ago and self driving or even remotely controlled cars still aren't really a thing?
Sorry, but it's massively easier to program something to fly within the confines of the NAS than it is within the road system where there are probably 1,000 more variables.
Click to expand...

No, it's easier to convince the public and the legislators to allow an aircraft to fly where it won't cost trillions of dollars to implement the infrastructure to make self driving cars actually work, and the public isn't adversely effected. You know, the whole cost/benefit thing our government worries about. It's a lot harder to pick your way through a set of cells than it is to pick your way through crawling traffic. Holes you decide to go through in an airplane can close up around you, and you HAVE to continue, possibly getting struck by lightening, getting pummeled by hail, or hitting wind sheer/microbursts that will have your airplane come out of the sky in pieces. When was the last time you saw hail on your radar? Or how about flying into un-forecast SLD? Holes you try to pick through in traffic can close up around you, and you can just stop and wait till it opens back up.
 
Yup, that's a lot harder to do. That's why airplanes crash at a much higher rate than cars.
 
esa17 said:
The flaw in your thinking is two fold.
Click to expand...

I actually agree with your point, I just don't agree with the timeframe you believe it will take place in.

esa17 said:
The point is, for those of us who know how it works, we know that we will see the single pilot airliner before my son is eligible for his ATP.
Click to expand...

I'm one of "us" who knows how it works, and it really all depends on how old your son is. My statement in my original post was that it wouldn't happen in any of our (as in, those of us who are all ready working as professional pilots) career-spans.
 
Last edited:
Computers cannot see, and they're catastrophically bad at differentiation of abstract criteria.

Elevators, which have been automated for decades, break all the time, and the state model for elevators and trams is incredibly simple.

Twitter, one of the most heavily used pieces of technological infrastructure on the planet, has continual hiccups, server errors, slowdowns, duplicate posts, etc... and it's a comparatively simple infrastructure.

The gold standard of production infrastructure is "five nines" ... that is 99.999% availability. Achieving that or better is very expensive, requiring multiple redundant systems, each capable of carrying the full capacity. 99.999% would be unacceptable for aviation—do the math.

The further down the tree of technology we go, the less reliable succeeding generations become. Nobody who runs a large-scale production system lives on the bleeding edge... for long.

The failure modes of large-scale automation cannot ever be guaranteed fail-safe.

We as a country stopped investing in infrastructure years ago, and unaddressed inflation is rendering even modest infrastructure projects unacceptably expensive. If we can't even truly modernize ATC, what on this planet can possibly convince you that we're in a position to have autonomous AI operations in our system?

Speaking of AI, there's no such thing yet, and research into true ML has been remarkably stagnant since the late 1980s. I can sense the simpering computer scientists out there wriggling like Milton from office space, wanting to disagree.

Go for it. Show me what you've got.

As for the rest of you, the more you sincerely discuss the concept of the inevitable and imminent autonomous airliners, the more credence you lend an incredible idea. Is it possible to have autonomous airliners? Yes, it's quite possible. Is it possible for it to happen soon, given our current technological, economic and social statuses? Only if we have a "go for broke" reason to band together as a society and do it, and even then you're looking 20-30 years out just due to the basic development lifecycle of complex things.

Single pilot airliners, ground-based control, etc? Possible, but stupid. It's hard to get much cheaper than having two pilots up front... and the synergy of human performance and automation technology, coupled with the redundancy of a second set of eyes, is hard to beat when performing efficiently.

Now about comments vis a vis the 'Singularity', yeah. Wake me up when we move to ipv6.

-Fox
 
It all comes down to $.

Air travel is still relatively safe. Airlines are making profits. I don't see any need for big changes anytime soon.
 
z987k said:
Which is why we developed widespread UAV use over a decade ago and self driving or even remotely controlled cars still aren't really a thing?
Sorry, but it's massively easier to program something to fly within the confines of the NAS than it is within the road system where there are probably 1,000 more variables. I mean the sensor array to figure out what is going on in the road system is going to be insane. You need almost no sensors to make a UAV.
From a programming and computer logic standpoint, flying in the IFR system is stupid easy in comparison to driving.
Click to expand...

You keep thinking about how easy things are when everything goes according to plan. Yes, an FMS and autoflight system can do a (pretty) good job of flying an airplane without pilot input when everything is going rosy. But most autoland systems can't even land on a single engine after you've got everything stable and trimmed for it. Hell, most autoland systems can't even land with a strong crosswind on two engines. And ask a 737 pilot how great the autoland is even when things are smooth and perfect.

Sorry, but computers are great at following magenta lines. That's about it. Throw in an engine failure at V1, and the technology is still a couple of decades away for the automation to handle that with a pilot sitting there keeping an eye on it. You guys thinking that we can hand over control of airliners to computers anytime in the next 30 years are living in a dream world and probably don't have much experience flying airliners.
 
Every time I think about installing automation for terrain avoidance or taking pilots 100% out of the flight deck I recall when I had every single terrain related alarm go off going into MAF
(flat and near sea level) at 4000ft. I can't help but wonder what the aircraft would have done if it had some sort of automated terrain escape feature and what a 100% automated aircraft would have done to resolve such an issue.
 
Maurus said:
Every time I think about installing automation for terrain avoidance or taking pilots 100% out of the flight deck I recall when I had every single terrain related alarm go off going into MAF
(flat and near sea level) at 4000ft. I can't help but wonder what the aircraft would have done if it had some sort of automated terrain escape feature and what a 100% automated aircraft would have done to resolve such an issue.
Click to expand...

I've had the GCWS in the F-15E go off at all kinds of altitudes nowhere near terrain.
 
mshunter said:
Read this as, "from my extremely biased opinion, it'll be here in the next 10 years."

No, it won't. Does the tech exist, sure it does. Will the public accept it, will the regulators allow it, and will the programmers be able to write code that satisfies everyone? Most likely not. It's more, much much more than just the capability to do it.
Click to expand...
How you should be reading it is: "Someone who know a lot more about this than I do thinks we've got about 22 years before the two pilot cockpit goes the way of the dodo."

But you go ahead and think what you want.

The instant accountants start figuring out its cheaper to use a datalink instead of a human the regulators will fall in line. The public will grumble and groan but they'll do what they have to so that they can save $3 on a fare.
 
SeanD said:
I think its far more important to develop technology that can clean my house and do laundry.
Click to expand...

Everyone went the girlfriend/S.O. route.

I was just thinking about plain old maid services. Low tech solutions are often best.

Acrofox said:
Now about comments vis a vis the 'Singularity', yeah. Wake me up when we move to ipv6.

-Fox
Click to expand...

I'm thinking of the hundreds of thousands of man-hours that went into teaching people the basic reasons/tenets of NAT, and realizing how we're going to fundamentally break most people's concept of IP addressing for good. It was a painful and eye-opening experience to learn to translate IPv4 addresses into binary. By hand. Ugh.

IPv6 is going to make heads explode worldwide.
 
esa17 said:
How you should be reading it is: "Someone who know a lot more about this than I do thinks we've got about 22 years before the two pilot cockpit goes the way of the dodo."

But you go ahead and think what you want.

The instant accountants start figuring out its cheaper to use a datalink instead of a human the regulators will fall in line. The public will grumble and groan but they'll do what they have to so that they can save $3 on a fare.
Click to expand...

Having down the line for two years now 121, and 3 years a mix of 91/135, I have come to understand a free things about this industry. The number one thing being that when people are in the thing, unless it's determined that blood spilt needs to cause the change, it happens very slowly. There currently is no airplane in development that is slated to be single pilot. That will take 20 years, minimum. So by that measure, once those come on line, the next logical step is to remove the pilot completely. Add another 20 years for that. So, 40 years, best case scenario. BEST case scenario. Not 10. Not 15. Not 20.
 
esa17 said:
How you should be reading it is: "Someone who know a lot more about this than I do thinks we've got about 22 years before the two pilot cockpit goes the way of the dodo."

But you go ahead and think what you want.

The instant accountants start figuring out its cheaper to use a datalink instead of a human the regulators will fall in line. The public will grumble and groan but they'll do what they have to so that they can save $3 on a fare.
Click to expand...

You clearly have very little understanding of how regulators work. Perhaps you should do a little research into how the railroads want to go down to a single engineer on their most advanced trains, but the rail regulators won't allow it. And you know what? The railroads are probably right that it's just as safe (unlike in this case). But the regulators don't care, because they want to see mountains of evidence proving it first. I give it 20 years until they'll approve even that. I'd give it at least 60 until you see a single pilot airliner, over a century before a zero pilot airliner is even a remote possibility.
 
You must log in or register to reply here.
Back
Top