Why creating autonomous aircraft should be a walk in the park.
I’m still heavily into podcasts, and one I listen to regularly is Freakonomics Radio. The quality of their shows are usually top notch, and while it feels like they are pulling more and more shows up from the archive for rebroadcasts, most of what they air is fresh (although a tad sensational and click-baity, from time to time).
On June 1 this year, they ran an episode called Why Does Everyone Hate Flying? And Other Questions Only a Pilot Can Answer. The episode features Patrick Smith, author of Cockpit Confidential, and the man behind the site Ask the Pilot. Patrick Smith might be Big in America, but I’d never heard of him, his book, or his site before he got interviewed on Freakonomics Radio, so good on him for getting some international exposure1.
One of the more interesting questions the show’s host, Stephen J. Dubner, asked Smith, is this:
I do know that autonomous cars are a potential reality. […] So talk to me for a moment about why I shouldn’t expect and fully demand all my planes be flown by robots and computers?
Autonomous aircraft means that there’s no need for pilot Smith in the cockpit, which in turn means that he’ll be demoted to serving drinks in the back of the plane. So he reacts as you’d expect to the question: He gets defensive, and pretty much dismiss people who even suggest such a outrageously crazy idea as people who have no idea what they are talking about.
I’m one of those people, and while I don’t have a “good grasp of the operational realities of commercial flying”, as Smith so correctly puts it, I’d still try to argue that making an autonomous aircraft is easier than making an autonomous car.
An autonomous aircraft’s simplified plan for getting from point A to point B would look something like this:
- Taxi from gate to runway
- Take off from runway, set course for destination, climb to cruising altitude
- Avoid bad weather en route
- Make course adjustments according to ATC
- At destination, enter correct glide slope towards designated runway
- Taxi to gate.
For all of the above, avoid other aircraft and UFOs.
An autonomous car’s simplified plan for getting from point A to point B would look something like this:
- Back out of the garage.
- Set course for destination
- Full stop at STOP signs
- Yield at intersections
- Stop at red lights
- Merge correctly on the high way
- Arrive at destination
For all of the above, avoid other vehicles, pedestrians, pot holes, aquaplaning, animals, adjust speed based on current road conditions and quality, slow down for road works, etc, etc…
Yeah, I have probably simplified the aircraft’s flight plan a bit, but my point is that there is a lot of stuff an autonomous car has to pay attention to that isn’t of any concern for an aircraft with an artificial intelligence at the steering column. Take other vehicles, for instance. Driven by inexperienced drivers, like me, and old grandmothers, like old grandmothers drive, they are a massive challenge - but it’s a challenge that will soon be solved.
So if building an autonomous aircraft is so much simpler than getting an autonomous car to behave in traffic2, why isn’t there Boeing and Airbus jets flying around with no pilots in the cockpit? First of all, the technology isn’t quite there yet. The autopilot feature of the Tesla S is perhaps the one that is most used on public roads these days, and it looks like it might have just killed a guy. Not on purpose, mind you, but because its sensors and emergency braking system didn’t react quickly enough to truck that turned in front of the Tesla.
Secondly, there is the massive shit storm that will hit the manufacturer of an autonomous aircraft with 300 people on board that falls out of the sky. Because that will happen. Most of today’s aircraft accidents are caused by human error, or at least a series of events with a weak human link in the chain of events. Removing the human from the equation won’t guarantee a safe flight, because you’ll still have the risk of mechanical failure. Also, the autonomous systems flying the aircraft will be created by humans. And humans fail.
So while it probably won’t happen in the next 10 years, I’m 100% sure we’ll see commercial autonomous aircraft in my life time, no matter how unlikely pilot Patrick Smith thinks it is.
vegard at vegard dot netwith your input. You can also use any of the other points of contact listed on the About page.
The way that you ultra-simplified the tasks, which you numbered 1 through 7, that an automatic airplane would be responsible for, shows that you have no grasp of just how much human input goes into getting a plane from A to B. Each of those seemingly simple steps that you’ve itemized can be further broken down into dozens of sub-steps, few of which are ever exactly the same, flight to flight. Flights are very organic, very fluid. Even the most automated cockpit can become very busy. The automation only does what we tell it to do: what, when, where, and HOW.
Please read here for more…
But I’ve got a pretty good idea about what it takes to get a car from A to B. You can break the simplified list of steps to get a car from A to B into several sub-steps that will be different for each trip, as with the ultra-simplified task list for the aircraft. Building an autonomous car is an incredibly difficult thing, but we’re just a few years away from this being a standard feature in every new car - at least in the high-end market - that’s manufactured.
Even though an autonomous aircraft is different, I can’t imagine (again, this is the layman talking) that it’s really that different. You’ve got an extra dimension, the Z-axis, to handle, and I the aircraft is a more complicated piece of machinery than a car, but it’s still just a matter of getting a thing from point A to B without anyone getting hurt.
Now, if you’re ever reading this, I’m curious to know how you managed to stumble across this teeny tiny site on the big, vast internet?