Will robots/AI replace human pilots?

FlyingTiger said:
I agree with the post above that cargo will be where it starts.
Click to expand...
Frederick W. Smith

Here’s a question from Helane Becker of Cowen. A lot of companies are starting to experiment with UAVs, FedEx experimenting as well. If so, how is that going and what do you think the timing would be of using UAVs both in the air and on the ground? Also, is it realistic to believe there could be unmanned trucks? Is FedEx going down this route (lol, pun intended)?

So Helane, obviously we at FedEx are very mindful of these trends. We have a number of activities underway in robotics in particular in the packaged handling sector. Henry and FedEx Ground are real leaders in this regard. In terms of UAVs in particular, we have five separate, I think it would be not fair to call all of them projects but work streams or projects in both aviation and automated vehicles. The difference with us and a lot of other people we’ll just prefer to keep working those issues and tell you about them when they make a meaningful difference in the company. I will say this much. I think our philosophy and we know a lot about these technologies. After all, our auto pilots in our 777 airplanes are among the most sophisticated robots in the world. They can take off, land the plane and taxi to the gate and turn themselves off if that’s what we chose to do so. (ed: Okay, that's a bit of an overstatement, but whatever) But it’s very difficult in the foreseeable future to substitute for the well trained pilot or driver or person. And we look at the use of automation more as an opportunity to improve the productivity of those types of experts within our system to make their job more comfortable and easy and above all to increase safety. So those five work streams are underway. You’ll hear a lot about them I’m confident in the next few years. But important in our philosophy maybe slightly different than a lot of other people that think that right over the horizon, everything is going to be an automated vehicle or some sort of UAV. We think that is unlikely and that this technology like most technologies particularly aviation technology will evolve incrementally over time with a great emphasis on safety first. So that’s our UAV update. We’ll take two questions now.

http://seekingalpha.com/article/400...-results-earnings-call-transcript?part=single
 
"Yes eventually" is the right answer to the question. (time frame unknown)

Cheers
 
For passenger flights, the day the engineer loads his entire family and his engineering team and all of their families on board and lets the system they built do the flying, is the day the public will start to trust it. And they'll still not fully trust it.

I think that particular but necessary PR stunt is still a large number of years away.
 
weilke said:
Three word answer: Wei Too Lo.

For every plane that was rescued by heroic crew intervention theere are two that got smashed up due to crew screwup.
Click to expand...
The reference to Asiana 214 raises an interesting question: Will it be possible to develop robots/AI that can fly visual approaches consistently and safely?
 
Sluggo63 said:
What happens when the bad guy takes over the ground station?
Click to expand...
Or the control link between the ground and the aircraft.

Are hackers going to stop getting better at what they do?
 
Palmpilot said:
Or the control link between the ground and the aircraft.

Are hackers going to stop getting better at what they do?
Click to expand...
Once we see automation on the ground, start looking to the skies. I think if we start having drones driving taxis, then it won't be a leap to think that same automation will be carried to other stuff.

I think the 777 automation quip wasn't entirely off base. From my understanding, most of the latest and greatest planes already do their thing mostly automatically... The pilot is just there to tell it what and when to do it and to take over if something goes wrong. Hell, I know on some flights Delta *insists* that the plane be allowed to self land without pilot intervention just to make sure the system is functioning properly. When you're mandating that the human not touch anything, eliminating humans is definitely in that particular development chain.
 
ircphoenix said:
I think the 777 automation quip wasn't entirely off base. From my understanding, most of the latest and greatest planes already do their thing mostly automatically... The pilot is just there to tell it what and when to do it and to take over if something goes wrong. Hell, I know on some flights Delta *insists* that the plane be allowed to self land without pilot intervention just to make sure the system is functioning properly. When you're mandating that the human not touch anything, eliminating humans is definitely in that particular development chain.
Click to expand...
That 777 quip is so far off base, it's in line at the concession stand trying to buy a $8 hotdog and $12 beer.
Our auto pilots in our 777 airplanes are among the most sophisticated robots in the world. They can take off, land the plane and taxi to the gate and turn themselves off if that’s what we chose to do so.
Click to expand...

I'll tell you as someone who has (up until very recently) been flying new 777's that:
1. They can not takeoff by themselves.
2. They can autoland (more on this later).
3. They can not taxi by themselves.
4. They can not shut themselves down.

Ok, autolandings... Your Delta Airlines anecdote is probably mostly true. When I started at my airline, each airplane's autoland system had to be checked every 30 (I think... I can't remember. Might have been 60) days. Every autoland that was completed was entered into the maintenance log as either SAT or UNSAT. A SAT autolanding would reset the 30 day clock, and an UNSAT would knock the plane into a CAT I-NO AUTOLAND status. Now days, maintenance can run a system test and check the autoland system and log it off that way.

The only time an autoland is now required for us is when the weather dictates that it is required that the plane autolands. We're talking low RVRs, 1000 down to 300 RVR. Other than when it's required for weather, I'd say 99% of the time, the plane is going to be landed manually.

I know I'm probably going way too in depth in the whole autoland system, but I'm trying to combat the autoland "myth" that is so prevalent when laymen and unfortunately, fellow pilots believe.

When I talk about autolands, I'm going to be talking about the full-up, have enough fidelity to land without the pilots seeing anything, 300RVR autoland. There are other modes where autoland is available, but the weather has to be good enough for the pilots to see to land, in case something squirrely happens at some point close to the ground. There are plenty of things that will kick a plane out of a full autoland status. For example, in the 777, which has 3 hydraulic systems, 3 autopilots, 3 ILS receivers and 3 radio altimeters you need all three of each for a full CAT III<600 RVR autoland. If you lose any one of those items, you now must see-to-land because the autoland system isn't reliable enough to assure a safe autoland. If you lose two of those systems, manual landing only (in fact, if you lose the left hydraulic system autoland is not available, even with the other two systems). How many more redundant systems are you going to put onto an airplane with the associated cost and weight to eliminate the pilot? Will you have 9 hydraulic systems? Nine ILS's?

All those issues are for aircraft status. You will also have to upgrade every commercial airport in the world to a full CAT III autoland airport. There aren't too many right now that will allow a 300RVR landing, and now you're talking about upgrading every piece of pavement that a commercial airliner may land at, the divert airfileds, ETOPS alternates, emergency airfileds, etc. That's a whole lot of CAT III ILS's with the monitoring equipment, flight checks, etc.

Lastly, becuase I'm getting tired of typing, this autoland stuff is great when the weather is low, which usually means clam winds and stable air. I'll tell you that autolands don't do as good of a job in bumpy, gusty winds. The crosswind limit in the 777 is 38 knots. If you're autolanding, the crosswind limit is 25 knots, but that's for visual conditions. If you are doing a CATII/III autoland, the limit drops down to 15 knots. Any more than that and it can't be assured that the autoland will conclude safely.

I'm sure that engineers are working on all the above issues, but when it comes down to it, it's going to come down to money. Will the airlines be willing to pay for all these upgrades, testing, etc. to eliminate the very efficient (and relatively cheap) backup that is the pilot. I don't know, but I think the answer is much more complicated (and expensive) than "we'll just let all the airliners autoland, they do it already..."
 
ya...it'll never happen.:confused:

btw...I was that guy who did the hazard analysis for Army UAS systems.
 
Art VanDelay said:
Yeah well I can tell you from experience that the autothrottles on the 777 ain't all that great. If you get into mountain wave or moderate to severe turbulence those autothrottles will launch you right on past Mmo !
Click to expand...
Same thing with the 737--even the newest ones right out of the factory. It can't--doesn't even try--to hold airspeed within 10kts on descent. In fact, it won't add power until you're 15kts below the set airspeed.

My favorite is when it's all setup for a VNAV descent. It has the latest winds uplink. It knows all the altitude and airspeed restrictions on the arrival. It gets to pick when to start the descent. So it calculates a the top-of-descent point and starts down. A few seconds after the thrust levers hit flight-idle a message pops up saying, "UNABLE NEXT ALTITUDE". Really?

CC268 said:
My buddy who flies for Southwest as a Captain always jokes about how easy it is to fly the 737 because the Boeing engineers did such a great job designing it
Click to expand...
Spoken like someone who's never flown an airliner other than the 737! LOL The airplane is handicapped by it being restricted to its 1960s design.

ircphoenix said:
I think the 777 automation quip wasn't entirely off base. From my understanding, most of the latest and greatest planes already do their thing mostly automatically.
Click to expand...
No more so than a car on cruise control is doing its thing mostly automatically. Just like the driver, the pilot is constantly there monitoring, reprogramming, verifying, and monitoring some more. The automation does the simple tasks very well. It doesn't do the complex tasks at all.
 
Checkout_my_Six said:
how does the military fly their +40,000 lb UAS?....amazing. :D
Click to expand...
Very expensively. First, those things crash at an outstandingly high rate. Much higher than any other MWS. And that's not combat losses, that's just flight losses. Second, launching and operating an RPA takes two to three times the manpower of a manned platform. Lastly, these things don't takeoff and land themselves. There is a "takeoff and landing" operator that has to be onsite (deployed) to the location. Once it's airborne, control gets passed to the operator in Nevada (or where ever).
 
weilke said:
10 years from now cargo jets will be single pilot with a autoland capable autopilot that can be remotely operated from the manufacturers emergency backup center. The system will have redundant datalinks through both satcom and terrestial stations at the major airports. 20 years from now new regional jets will be single pilot using the same by then proven, technology. At that point, parcel freight will go unmanned.
Click to expand...

These types of posts always make me laugh. No offense weilke but making a blanket statement as if you just hopped out of your DeLorean and know this as absolute fact always puts a smile on my face.

I think there is a chance we'll see some fully automated aircraft, but I don't think it will be anytime soon. As other people have said, there's just too much gap in where technology is now (where it can do the job if everything goes right) to where it'd have to be (nearing 100% reliability regardless of weather, solar flares, winds, NOTAM'd out approaches, etc) - hell there might even be people out there flying around with no radio or transponder! OMG, how will the robot plane miss that one? Just too many what ifs, my guess is we are at least a generation if not two generations away from automated passenger carriers. Time will tell, I certainly think that will be the goal for businesses - maybe less so for passengers.
 
or....God forbid ....a manned aircraft fly into it. :eek:

One thing is for sure.....the technology will enable single pilot 121 ops.
 
Checkout_my_Six said:
or....God forbid ....a manned aircraft fly into it. :eek:

One thing is for sure.....the technology will enable single pilot 121 ops.
Click to expand...
Make sure you reconfigure all those airplanes to have a lav in the cockpit - otherwise you'd end up with a flight attendant only up front when we need to take a leak. Also right now we have to have the face-sucker O2 mask on anytime there isn't two pilots on the flight deck so how do I eat my lunch? How do they hear me talk on the radio? What is my max flight time with all that extra gear on? How much more are you going to have to pay me to fly by myself with no one to talk to all freakin day? What happens when joe the solo pilot falls asleep because he's not had anyone to talk to him all the way across the Atlantic? While you are reconfiguring the cockpit for the lav make sure you move all the circuit breakers, reference manuals, override panels, relays, radios and printers to a position that I can reach them without getting out of my seat, but also increase the visibility for taxi ops since I can't see around the corner of my big airplane.

Not to mention that we always have to have two people on the flight deck at a time due to a (very unlikely but possible) medical issue or a GermanWings type situation.

Still think it's a stretch... a big stretch.
 
Yes....if you dig into the causality of those you'll notice a lack of discipline and human factors involved.

Takeoff and landing incidents do not dominate that data.

Much has changed since the majority of those incidents were documented. Also, consider the war time environment. Not quite the optempo nor the level of operational risk here at home.
 
Last edited:
EvilEagle said:
Make sure you reconfigure all those airplanes to have a lav in the cockpit - otherwise you'd end up with a flight attendant only up front when we need to take a leak. Also right now we have to have the face-sucker O2 mask on anytime there isn't two pilots on the flight deck so how do I eat my lunch? How do they hear me talk on the radio? What is my max flight time with all that extra gear on? How much more are you going to have to pay me to fly by myself with no one to talk to all freakin day? What happens when joe the solo pilot falls asleep because he's not had anyone to talk to him all the way across the Atlantic?
Click to expand...

Why would you want to fly across the atlantic if all you got to do is to ship 12 tons of parcels from Memphis to Des Moines at 3:30 in the morning ? This wouldn't be the first single pilot operation in the history of aviation.

Not to mention that we always have to have two people on the flight deck at a time due to a (very unlikely but possible) medical issue or a GermanWings type situation.
Click to expand...

This thread was about AI and automation. In either of those scenarios the plane will either land autonomously or by remote control. Or maybe it crashes.
 
weilke said:
Why would you want to fly across the atlantic if all you got to do is to ship 12 tons of parcels from Memphis to Des Moines at 3:30 in the morning ? This wouldn't be the first single pilot operation in the history of aviation.
Click to expand...

I'm very familiar with single pilot operations. They are great when you are fighting, they suck a little more when you are tired or slightly distracted by other parts of life.


weilke said:
This thread was about AI and automation. In either of those scenarios the plane will either land autonomously or by remote control. Or maybe it crashes.
Click to expand...
That's a big maybe.
 
Have a friend in the industry who is a little contrarian - someone told him the "last fighter pilot has already been born". His rebuttal was the "last drone pilot has already been born", that AI will handle the flying for high-end drones, other than the toys and hobby stuff. Still gonna be human(s) in the airline (and military) cockpits for a long, long time.
 
My question is, at what point do people designing these kinds of things realize that all they are doing is taking away jobs from people? At this rate, in 20 years everything will be done by machines, truck drivers will be out of work, taxi drivers out of work, and eventually pilots... I guarantee half of Americans could not get from New York City to Miami without a GPS despite the simple fact the 95 takes you all the way there. Cars that park themselves, hell even drive themselves. Technology and automation are great to some extent, but it's getting to the point where the smarter machines we have, the dumber and less skilled people are, to me that is the sad part. If I have kids one day they will probably never even know what a manual transmission is. Machines are taking half the fun out of life, I became a pilot so I could fly airplanes, not ride in one. Ok, rant over.
 
just think....we could all still be riding horses....if it weren't for Henry Ford. :eek:
 
Checkout_my_Six said:
No they don't. How much UAS experience do you have?
Click to expand...
Me, personally? None. Have some good friends flying RPAs in ANG units. They tell the same tale. So do the safety briefings that we get during drill weekends. These things are not as reliable as people think. They crash at a greater rate than any other AF MWS, and that's only the ones that the AF reports.

Checkout_my_Six said:
One thing is for sure.....the technology will enable single pilot 121 ops.
Click to expand...
Not for a very, very long time. How much large, Part 121 experience do you have?

Checkout_my_Six said:
Yes....if you dig into the causality of those you'll notice a lack of discipline and human factors involved.

Takeoff and landing incidents do not dominate that data.

Much has changed since the majority of those incidents were documented. Also, consider the war time environment. Not quite the optempo nor the level of operational risk here at home.
Click to expand...
Again, I agree. These are not takeoff and landing accidents. These are in-flight, the drone goes stupid and crashes.

I don't know how much has changed since when. Here's an article from January of this year. Ten months ago.
https://www.washingtonpost.com/news...re-crashing-than-ever-as-new-problems-emerge/

I get it. You work in the RPA world and think that UASs are the panacea for all the ills of manned flight. I just don't buy it. I don't know much about the RPA world. Just what I read, get briefed on, and hear from my buds over beers. What I do know about is what goes on day in and day out in the cockpit of one of the world's newest, most technologically advanced aircraft working in Part 121 operations. And the technology I see is a far cry from even being close to single pilot operations, let alone pilotless. I could be wrong.
 
No...I did work for the UAS world (Textron Army UAS aviation) for almost a decade....

btw the only 121 experience I have is rolling a 777 simulator....then landing it.:D
 
Last edited:
Sluggo63 said:
...And the technology I see is a far cry from even being close to single pilot operations, let alone pilotless. I could be wrong.
Click to expand...
Without knowing your background I'll say that the average line pilot, military or civilian, may not be in touch with how fast the state of the art is advancing. Autonomy (not just remote piloting) is making leaps and bounds now and what's fielded is not representative of what's probable/possible in the not-too-distant future. Look at the X-47 (autonomous, not remote-piloted) to see the advancement (in terms of operational capability) that's been made in the past 5 years.

I don't think single-pilot 121 ops are just around the corner but I think it's inevitable...I also doubt that technology is the largest hurdle to overcome ;)

Nauga,
who sometimes doesn't even get a vote
 
10 years ago we weren't discussing drones other than to say "hey, these predators are launching a single missile at a time. Isn't that interesting?"

Now we have the X-47, RQ-170 and God knows what else. Hell, for a thousand bucks you can go buy a GPS stabilized pocket drone with a 4k camera, ability to beam an image back to you from 3 miles away, travel at 40 mph, and autonomously return to its point of origin or fly a predetermined pattern around a user defined object without a wristband or other device to lock in on.

10 years ago we were talking about the Tesla roadster, and how Tesla was going to be defunct by the end of the year. Now we're talking about their autopilot system. Google hadn't even started their driverless car program yet... and now their vehicles have done 1,000,000+ miles of autonomous driving.

I think we're still a ways off from completely dropping humans from the loop... but not quite as far as some may think.
 
nauga said:
I don't think single-pilot 121 ops are just around the corner but I think it's inevitable...
Click to expand...
In order for it to be inevitable, it has to be advantageous. What do you expect the advantages to be, and how wiil they outweigh the disadvantages?
 
Palmpilot said:
In order for it to be inevitable, it has to be advantageous. What do you expect the advantages to be, and how do you expect them to overcome the disadvantages?
Click to expand...
Increased reliability....with reduced risk. That'll be the advantage of the automation.....once single pilot ops proves it out. :)
 
Probably never gonna be all-or-nothing, either way. Video conferencing doesn't replace all face-to-face interaction, it just cuts back on some travel. I can imagine single pilot scheduled ops, but probably not pilotless scheduled ops for several generations (of people, not tech). The recent advances in tech are great, but are applied science, using the fundamentals in HW and SW that have been around a good while. Gonna need a new, radical change, in SW, ground up, for common commercial use involving safety of life without human oversight.
 
nauga said:
I don't think single-pilot 121 ops are just around the corner but I think it's inevitable...I also doubt that technology is the largest hurdle to overcome ;)
Click to expand...

Even if the FAA signs off on it, the unions never will. Got to have that stoker on the electric train engine in case some coal needs shoveling.
 
That's equivalent to saying it will be successful if it's successful.
 
Larry in TN said:
Same thing with the 737--even the newest ones right out of the factory. It can't--doesn't even try--to hold airspeed within 10kts on descent. In fact, it won't add power until you're 15kts below the set airspeed.

My favorite is when it's all setup for a VNAV descent. It has the latest winds uplink. It knows all the altitude and airspeed restrictions on the arrival. It gets to pick when to start the descent. So it calculates a the top-of-descent point and starts down. A few seconds after the thrust levers hit flight-idle a message pops up saying, "UNABLE NEXT ALTITUDE". Really?


Spoken like someone who's never flown an airliner other than the 737! LOL The airplane is handicapped by it being restricted to its 1960s design.


No more so than a car on cruise control is doing its thing mostly automatically. Just like the driver, the pilot is constantly there monitoring, reprogramming, verifying, and monitoring some more. The automation does the simple tasks very well. It doesn't do the complex tasks at all.
Click to expand...

LOL....
 
ircphoenix said:
10 years ago we weren't discussing drones other than to say "hey, these predators are launching a single missile at a time. Isn't that interesting?"

Now we have the X-47, RQ-170 and God knows what else. Hell, for a thousand bucks you can go buy a GPS stabilized pocket drone with a 4k camera, ability to beam an image back to you from 3 miles away, travel at 40 mph, and autonomously return to its point of origin or fly a predetermined pattern around a user defined object without a wristband or other device to lock in on.

10 years ago we were talking about the Tesla roadster, and how Tesla was going to be defunct by the end of the year. Now we're talking about their autopilot system. Google hadn't even started their driverless car program yet... and now their vehicles have done 1,000,000+ miles of autonomous driving.

I think we're still a ways off from completely dropping humans from the loop... but not quite as far as some may think.
Click to expand...

A lot further than you seem to think....

Google has done a million miles and has an accident rate completely consistent manned vehicles over the same miles. It's not zero. But it is very small; a million miles isn't that much for this purpose.

If you have ever shared the roads with one of these vehicles, you simply will not believe the hype. There is no way on God's green earth a vehicle that jams on the brakes when two people stand on a corner having a conversation improves safety, congestion, or anything else. These vehicles drive like stoned teenagers or scared little old ladies. They are NOT good drivers.

And they don't work in the rain. Oops.

They are not close to unassisted operation. That's years off, at least. Maybe a few assisted operations.

You won't do well trying to predict technology by assuming it's magic. PopSci has tried that over its entire lifetime. Go pick up a back issue from the 70s and see what it thought we might be doing now. You have to look deeper, and closer in. Maybe someday "AI" might be ready, but it's not close just because it can handle perfect conditions. It's the off-nominals that determine when a safety critical system works. That's FAR away.
 
You must log in or register to reply here.
Back
Top