Unfrozen caveman B-52 driver here. Lost while on a Chrome Dome mission in 1957, I was trapped in the Greenland ice pack until global warming thawed me. Did you guys know it’s getting warm up there?
I’ve come back from the frozen wastes to take a look at the future of aviation.
Keep in mind that my view of technology is stuck in the 1950s. Say, what do you think about those new "transistors"? Pretty keen huh? And have you seen the fins on the new DeSotos? Wow!
Let’s talk about those new-fangled drones you kids are all playing with these days. Drones have actually been around a lot longer than you think. Unmanned aircraft were around in experimental form as far back as WWI. The Ryan Firebee first flew in the early 1950s and was still in use as late as 2003.
Now the difference is that the early drones were little more than radio-controlled aircraft. Modern drones are much more autonomous.
The success of commercial drones in recent years leads the “Let’s automate everything (except my software developer job)” bunch to propose pilotless airliners. Usually they want to start with cargo aircraft, because they know most people with a lick of sense won’t get on a pilotless airplane, at least for now.
Now having some experience with cargo operations, the flaw here is that cargo aircraft operate from the same airports, runways and airspace as passenger aircraft. If your unmanned cargo plane collides with a planeload of passengers the end result is the same. Likewise if it crashes into a suburban neighborhood.
Keep in mind that the military has crashed a fuckton of drones over the last decade or so. I’m told that those don’t count, however, because reasons. In 2015 the Air Force lost ten Reaper drones, at a cost of $14 million a pop due to electrical failures. To date roughly half the Predator and Reaper drones built have been lost. The loss rate for drones has come down significantly, however so has the loss rate for manned aircraft.
The MQ-1 now has a loss rate pretty close to an F-16. Mind you the nickname for the F-16 is “lawn dart” so that bar has been set pretty low. The MQ-1 also costs almost as much as an F-16 and carries a lot less ordnance. Its main advantage is that it can loiter for a very, very long time. This makes them great for counter-terrorism, but against a “real” enemy they would last about 5 minutes. If I was an enemy the first thing I would do is go after the satellites, because I know just how dependent on them we’ve become.
Now modern autoflight systems are pretty awesome. I have little doubt that with a few modifications you could probably tow an A380 or 777 out to the runway, send it off to Tokyo and have it land itself. Assuming that the entire world got out of its way, and nothing broke, and it didn’t fly itself through a thunderstorm, and there was a working CAT III ILS at the other end.
Since I’m most familiar with the 757/767 I can tell you that there are a lot of things that can prevent it from being able to autoland. At least two of the three autopilots need to be working. All three hydraulic systems need to be functional. Two out of three generators need to be online. The 757 needs to have both engines operating, the 767 can actually do a single-engine autoland. At least two of the ILS receivers need to be working.
Most importantly, the runway needs to have a functioning ILS transmitter plus it has to be accurate enough for autoland operations (not all of them are).
There is a newer system called GBAS, that allows for autoland from a GPS approach, but it still requires a ground-based transmitter at the airport to correct the raw GPS data. This is called a GLS approach, and not many airports have them (yet). Presumably an autonomous aircraft would be limited to certain airports or it would need to have sensors so awesome that it could perform a purely “visual” approach.
I think the real test of automation would be compound emergencies. These are the ones that make headlines like Sully landing in the Hudson, United 232 at Sioux City or Qantas 39.
Qantas 39 is an interesting example. They had an “uncontained” engine failure (basically it blew up). When the engine went it damaged the nacelle, wing, fuel system, landing gear, flight controls, the controls for the other engine on that wing and caused a fire in the left inner wing fuel tank. That’s about as bad as it gets folks. This generated something like 100 ECAM checklists, some of them conflicting, that the crew had to work through. They were lucky in that they had not two but five pilots in the cockpit. One was an extra for the long flight and the other two were highly experienced Check Airmen.
In a worst case scenario, let’s say loss of all AC electrical power, caveman here can land the jet. Can’t say I’d want to do it but that’s what they pay me for. An autonomous airliner will need to have that kind of capability.
Even then, the real stumbling block isn’t flight control software so much as our air traffic control system. Currently our ATC system depends on radar tracking of aircraft and voice control over VHF radio. I’m amazed it works as well as it does for how old it is. The longer I do this job the more convinced I am that Air Traffic Controllers are magicians. If you want to hear true wizardry in action, listen to New York Approach when they’re working a stream of arrivals into JFK or Newark.
Unmanned aircraft would have to be able to respond to air traffic control commands somehow, presumably via data-link. Preferably with triple redundancy, because anything important in aviation has at least one backup and preferably two. As I was taught back in Flight Engineer training “Mister Boeing doesn’t like a single point of failure”.
That data link also has to be resistant to jamming, spoofing or hacking. That may sound like something out of a James Bond movie, but we think that’s how the Iranians got their hands on one of our advanced military drones. As airliners become more “connected” via inflight entertainment systems we’re starting to see hackers targeting them. Hopefully there’s a “firewall” between the entertainment and flight management systems.
A lot of what we do as pilots is science, but a good part if it is “art”, based on experience and judgement. Flying an airliner sounds easy enough when you think about it. After all, we’re just going from Point A to Point B. It’s dealing with the complexity of the weather and very busy airspace in and around major airports that’s challenging. I’ll freely admit that my job can be pretty boring 80 percent of the time. It’s the other 20 percent where I make my money. Even with all the automation engaged, two pilots can have their hands full on a busy arrival into JFK or LAX. Just taxiing at some of these airports can be task saturating “Give way to the United 737 on Alpha, then taxi to 28 Right via Foxtrot, Bravo, Bravo One, hold short of 28 Left, contact tower on 121.8”
For example, just the other night we had to pick our way around, between and over a massive line of thunderstorms that stretched across much of the country. It’s times like that I wish I could be sitting on the ground somewhere flying the thing remotely.
This involved both the Captain and myself interpreting the airborne weather radar plus what we can see with the Heads Up Display and FLIR (Night Vision). We also talked to our dispatcher to find out what the ground-based radar was showing. Finally we coordinated with ATC to find out what their radar was showing and how everyone else was dealing with this wall of weather stretching from Houston to St. Louis.
I have yet to see an autopilot that can interpret radar returns from a heavy storm system, talk to ATC to get a feel for what THEY are seeing, and make a sound decision. Even with a very good weather radar, we still use the old fashioned Mark I eyeball to look at a cumulus buildup and go “Nope! Not going in there!”
Likewise the autopilot doesn’t know that New York Approach is probably going to keep you up high going into Newark until you get over Teterboro and then “slam dunk” you with a tight right turn and rapid descent to 22 Right. But you damn well better be ready for it because you’ll never get slowed down in time otherwise. I know, silly me and my real-world examples.
IF Airport = Newark AND Runway = 22R AND Clearance = “180 to the Marker” AND Tailwind > 10 knots THEN
You = Screwed
END
So I think we may have a few years before fully pilotless airliners are roaming the skies. I don’t think the technology is mature enough and mainly the Air Traffic Control infrastructure doesn’t exist for it. Will I see it in my lifetime? I’d say that depends on how long I get to stick around.
What about semi-autonomous? I’m sure that’s coming. There’s a very old joke about the cockpit of the future containing a single pilot and a large dog. The pilot’s job will be to feed the dog. The dog’s job will be to bite the pilot if he tries to touch anything.
The one that worries me the most is being cooked up by our friends at DARPA. Yep, the same folks that gave us the internet are trying to free us from the drudgery of flying (and collecting paychecks).
ALIAS as it’s called is meant to be a “drop in” replacement for a human copilot. The idea being that instead of two pilots in the cockpit, there would be one pilot systems monitor with R2D2 here actually flying the jet. “Through ALIAS, the pilot flies the aircraft by means of a tablet computer that recognizes familiar gestures such as swiping and tapping”.
I don’t think R2D2 would be much fun on layovers. If I have to fly with a robot, can I least get this guy?
What’s amazing about ALIAS is it actually learns how to fly different types of airplanes. Which also means it could probably learn to replace a water pump on a Chevy, perform cataract surgery, flip a hamburger, or file a legal brief. Just wait until it learns how to shoot a gun, then hilarity ensues.
Right now ALIAS is being developed for the military, but I could see it making its way into the civilian world in a few years. Theoretically an airline could replace half their crew force with these robotic “copilots”. Thanks DARPA!
So how long before this all happens? Hard to say. The technology is advancing very rapidly, but there is a lot of inertia in government agencies, corporate boardrooms and public opinion. The certification process for any new airliner is lengthy and expensive. I would think it would be even more so for the first completely unmanned airliner. Not even automation-crazy Airbus has a pilotless airliner currently in development, at least that I know of.
People are rightly going to be a bit wary at first. 300,000 pounds of robotic 767 landing in your back yard is going to make a bigger mess than the cute little plastic quadcopters we’re all familiar with.
I’m hoping to make it to retirement age before any of this comes to pass (because it’s all about me). I can retire in 5-10 years, so I’d say that’s a pretty safe bet.
What I would expect to see first is single-pilot operations, with a robotic “copilot” like our friend ALIAS or possibly a ground-based pilot as a backup. Once that becomes accepted I would expect to someday a single pilot on the ground monitoring one or more largely autonomous aircraft. The only way taking the pilot completely out of the aircraft remotely makes financial sense is if they can manage more than one aircraft at a time.
Keep in mind that this would require a massive investment in infrastructure, especially the air traffic control system. The amount of satellite bandwidth required would be very expensive. The military has that kind of money to throw around but corporate bean-counters can be notoriously tightfisted.
I don’t think this will affect old-timers like myself. I do worry about anyone who is just now entering the industry. Buckle up, the ride’s going to get bumpy.