An emergency landing system that small aircraft passengers can activate | Hacker News Reader

An emergency landing system that small aircraft passengers can activate

economist.com

9 pointsjkuria11 comments

At this point I think it is fair to say that pilots are only kept around for passenger comfort. Automated aircraft have been in the skies since the 1940s and as computers became more powerful we have reached a level in which flying without computer aides is not only difficult, but literally impossible for a many new models (aircraft as far back as the F-16 in the '80s are impossible to fly straight by hand).

Given the strict and well defined procedures regarding commercial air travel, I see it as only a manner of time until air traffic controllers are giving commands with keyboards instead of their mouths. That is, until automated air traffic control completes the circle.

Automated air travel is, by almost all accounts, a solved problem. None of the ambiguity that plagues self driving vehicles exists in the air. Not only does the plane have basic location (INS + GPS) and attitude knowledge, but it also has access to synthetic terrain databases and live weather radar combined with hundreds of ground radar stations offering NDB and VOR/DME services, and to top it all off most commercial runways have ILS installations that can bring the airplane within inches of the centerline and perfect glideslope.

It is a bit of a shame as someone who has wanted to be a pilot for a long time. I still want to get my PPL, and I wholeheartedly believe GA will be alive for decades to come even after the Boeing drones come out, but at this point I can't see it as being a valid career option.

Change my mind?

Autopilot can't deal with unexpected situations or instrumentation failures particularly well. Take the Sully landing for example - without specifically setting up parameters for that particular case, how could autopilot save all passengers on board?

Well, it is called the miracle on the Hudson for a reason. Software is only as capable as the hardware it is running on. If you pulled the brakes off a Model 3 and brake-checked it, you'd be hard pressed to find a human or Autopilot version that could solve that problem. In special cases, you might have a lane open to swerve into - like Sully did with the Hudson - but such maneuvers are rarely available. To be fair, dual engine loss on takeoff is also not really a common occurrence.

Truthfully, Sully's incident represents the peak of human ingenuity, and I concede that it will take us decades if not centuries to reach a level of automation that can match that. But an autopilot that can react appropriately to 99.999..% of cases? I think it could already exist. Certainly, if dual engine loss is on roadmap, proper software could be developed to ensure the best possible scenario. I think that it's never been a consideration for the industry because these incidents are so rare, and also because that effort is better spent in trying to build more reliable engines. An ounce of prevention is worth a pound of cure and all that.

If AI can be cheaper than human pilots, then the cost savings can easily be thrown into massive parachutes for cabins like those in smaller private planes. That, or solid state emergency landing thrusters.

Previous discussion:

https://news.ycombinator.com/item?id=21407727

Vendor page:

https://www.garmin.com/en-US/autonomi/

One might think this kind of technology will become particularly important if we get to the point of having self-flying planes.

I mean, should this technology actually be very reliable in adverse conditions, it’s 50% of a self-flying plane. I don’t know how much auto-landing currently exists in the commercial airliner industry, but I suspect it’ll be much easier to build self-flying planes than self-driving cars.

Autoland has existed in commercial airliners since the late '80s. It is really not much more than a mathematical equation given the feedback provided by ILS (which tells the airplane exactly how far off it is from the runway in both lateral and vertical directions).

However, pilots prefer to land by hand mainly because it gives them something to do (and it's fun). Additionally, not much thought was given to comfort when these autoland systems were made, and last I heard nobody recommends actually using them as they have a nasty habit of planting the plane into the runway (apparently, they're not programmed to flare before touchdown, resulting in all of the landing gear hitting the pavement at once). This is not really dangerous, but it does impact the lifespan of the landing gear and also gives the passengers quite a kick.

These problems are relatively minor and can be fixed in software, but we still have planes kicking from the '80s, so it will take a while for those changes to trickle down. Nonetheless, I am confident we can build a comfortable autoland in the present day.

Having played many flight simulators I fantasize about landing an aircraft in a emergency situation.

Previous submission/discussion (without paywall): https://www.piper.com/press-releases/piper-announces-new-m60...

Interesting topic but a paywalled article.

Reply on news.ycombinator.com