From the article: This was at Rocky Mountain Metropolitan Airport, a smaller airport on the way to Boulder. ... ...
From the article:
A Beechcraft King Air executed a safe landing in Denver under Garmin Autoland control on December 20, possibly the first use of the system outside of testing and certification—though the nature of the onboard emergency declared by the computer remained unclear [...]
Garmin's Autoland, the first certified system designed to land an aircraft without human input in cases of emergency, earned the 2020 Robert J. Collier Trophy. Part of the Autonomi suite of safety technologies, Autoland is designed to take full control if activated, and to do so automatically if the pilot becomes unresponsive, such as in cases of hypoxia.
...
The system, as designed, made additional calls as it flew the aircraft in a descending circle a few miles from the runway. Controllers advised various aircraft and ground units of the incoming emergency, and that rescue vehicles were maneuvering into position. About a minute before landing, the tower broadcast to the King Air, "If you can hear me, any runway, cleared to land, wind three-five-zero at six, altimeter three-zero-zero-zero."
Controllers told other aircraft prior to the landing that they expected the King Air to stop on the runway after landing and shut down, which it apparently did, as designed. Aircraft on the frequency after the King Air's landing were advised the airport was closed and was expected to remain so for at least 30 to 60 minutes while emergency crews responded.
What an absolutely fascinating piece of technology. I've seen so many videos of staging runs for autoland systems, or "live fire" tests, but I've never seen it used in anger like this, and it...
What an absolutely fascinating piece of technology. I've seen so many videos of staging runs for autoland systems, or "live fire" tests, but I've never seen it used in anger like this, and it performed brilliantly.
With all these amazing new super simple cockpit configurations coming out (like the Cirrus Vision) that take out a lot of workload off the pilot's back, aviation might just become more pilot-friendly than ever, superb work by all engineers that took part in all avionics/systems involved.
Just saw the VASAviation video on this earlier today. Interesting to hear it all. And really neat that we have this technology that can be implemented on smaller planes! While I'm sure these are...
Just saw the VASAviation video on this earlier today. Interesting to hear it all. And really neat that we have this technology that can be implemented on smaller planes! While I'm sure these are pricey now, maybe one day it'll be cheap enough to be standard on all planes.
Does this system communicate with air traffic control? What stands between today’s implementation and full self-piloting from terminal to terminal? I don’t know anything about piloting (and flight...
Does this system communicate with air traffic control? What stands between today’s implementation and full self-piloting from terminal to terminal?
I don’t know anything about piloting (and flight deck instrument panels are horrifyingly complex so I know there’s a lot to account for there) but it’s always seemed to me that autonomous passenger aircraft are likely an easier nut to crack than self-driving cars. Mainly due to the unpredictability of road conditions due to construction, traffic, wildlife, pedestrians, other obstructive hazards, weird one-off exceptions to normal rules of the road that require reading signs and improvising, nonverbal driver-to-driver communication, and so on. Ground vehicles are insanely hard to automate. Yet progress is being made there. I’ve always heard that takeoff and landing are the hardest things about flying a plane, and it sounds like this tech has half of that taken care of. Makes me wonder if airline pilots’ days are numbered.
Much like self driving cars, 80% of flight time is pretty easily automated. Autopilots have been around for over a hundred years, and in widespread use for almost that long also. A fully automated...
t’s always seemed to me that autonomous passenger aircraft are likely an easier nut to crack than self-driving cars.
Much like self driving cars, 80% of flight time is pretty easily automated. Autopilots have been around for over a hundred years, and in widespread use for almost that long also. A fully automated transatlantic flight, including landing and takeoff, was first made in the 40s.
The issue is that the last 20% is pretty difficult. The last 1% is almost impossibly difficult.
Flying is mostly boring until it isn't, and modern aircraft have literally tens of thousands of things that can go wrong with them, which right now require an experienced and capable human being to think on their feet to solve.
An autopilot to fully replace a human would need a human pilots experience, and the human ability to visualize complex, novel situations and think on their feet to solve them.
We're already there for situations where the risk to human life is smaller when a human isn't onboard, like when operating large military drones from bases and remote airfields in warzones. We're not there for operating large cargo or passenger jets, and for the latter, I'm kind of doubtful we ever will be.
An airline is responsible for the lives of half a thousand people on a transcontinental flight. It would be grossly negligent to not have experienced pilots on board to monitor the plane, even if you had utmost faith in autopilots. Besides that, crew costs aren't even in the top 10 of the costs airlines have to pay to get a plane in the air. Even ignoring the safety implications, eliminating pilots wouldn't be a significant cost savings for the airline when weighed against the negative public sentiment that would come with paying for a ticket on an uncrewed drone.
For small, personal air transport, yes I could definitely see autopilots taking over the whole job within the next 25 years. For airliners, I would be shocked if it ever happened. The risk calculus is totally different for the two.
About five minutes later, a computer-generated voice announced the King Air's call sign on the tower frequency and reported its position south of the airport, stated, "pilot incapacitation," and the system's intention to conduct an "emergency Autoland" on Runway 30R in 19 minutes.
From the article:
This was at Rocky Mountain Metropolitan Airport, a smaller airport on the way to Boulder.
...
...
What an absolutely fascinating piece of technology. I've seen so many videos of staging runs for autoland systems, or "live fire" tests, but I've never seen it used in anger like this, and it performed brilliantly.
With all these amazing new super simple cockpit configurations coming out (like the Cirrus Vision) that take out a lot of workload off the pilot's back, aviation might just become more pilot-friendly than ever, superb work by all engineers that took part in all avionics/systems involved.
Just saw the VASAviation video on this earlier today. Interesting to hear it all. And really neat that we have this technology that can be implemented on smaller planes! While I'm sure these are pricey now, maybe one day it'll be cheap enough to be standard on all planes.
Also some bonus ATC radio traffic of a test of one of these systems earlier this year.
Does this system communicate with air traffic control? What stands between today’s implementation and full self-piloting from terminal to terminal?
I don’t know anything about piloting (and flight deck instrument panels are horrifyingly complex so I know there’s a lot to account for there) but it’s always seemed to me that autonomous passenger aircraft are likely an easier nut to crack than self-driving cars. Mainly due to the unpredictability of road conditions due to construction, traffic, wildlife, pedestrians, other obstructive hazards, weird one-off exceptions to normal rules of the road that require reading signs and improvising, nonverbal driver-to-driver communication, and so on. Ground vehicles are insanely hard to automate. Yet progress is being made there. I’ve always heard that takeoff and landing are the hardest things about flying a plane, and it sounds like this tech has half of that taken care of. Makes me wonder if airline pilots’ days are numbered.
Much like self driving cars, 80% of flight time is pretty easily automated. Autopilots have been around for over a hundred years, and in widespread use for almost that long also. A fully automated transatlantic flight, including landing and takeoff, was first made in the 40s.
The issue is that the last 20% is pretty difficult. The last 1% is almost impossibly difficult.
Flying is mostly boring until it isn't, and modern aircraft have literally tens of thousands of things that can go wrong with them, which right now require an experienced and capable human being to think on their feet to solve.
An autopilot to fully replace a human would need a human pilots experience, and the human ability to visualize complex, novel situations and think on their feet to solve them.
We're already there for situations where the risk to human life is smaller when a human isn't onboard, like when operating large military drones from bases and remote airfields in warzones. We're not there for operating large cargo or passenger jets, and for the latter, I'm kind of doubtful we ever will be.
An airline is responsible for the lives of half a thousand people on a transcontinental flight. It would be grossly negligent to not have experienced pilots on board to monitor the plane, even if you had utmost faith in autopilots. Besides that, crew costs aren't even in the top 10 of the costs airlines have to pay to get a plane in the air. Even ignoring the safety implications, eliminating pilots wouldn't be a significant cost savings for the airline when weighed against the negative public sentiment that would come with paying for a ticket on an uncrewed drone.
For small, personal air transport, yes I could definitely see autopilots taking over the whole job within the next 25 years. For airliners, I would be shocked if it ever happened. The risk calculus is totally different for the two.