Improving flight control automation is useful, and relatively straightforward. The hard part is emergency procedures. If an engine fails then should the system attempt to compensate with the other engines, or trigger the ballistic recovery parachute? Does the optimal response change if you're at 50ft altitude instead of 500ft?
The general problem with relying on automation is that software developers can't necessarily anticipate and code for every possible failure mode. Whereas an experienced human pilot at least has a chance to invent new solutions in real time based on knowledge of first principles. For a prime example look at how Chesley "Sully" Sullenberger intentionally disregarded parts of the official checklist during the US Airways Flight 1549 emergency because he recognized that the procedure hadn't been written with their situation in mind.
It's important to keep in mind the average (or median) case rather than outliers. I.e. how frequently would an automated system outperform a human during an emergency.
Presumably the checklists were created for a good reason. Had Sully's outcome been different, there would've been a lot of hemming & hawing about not following procedure.
This is pretty much impossible to measure. If we could predict every failure mode imaginable, then I think it might be possible if you could predict humans. However, humans are not predictable and you can’t predict every failure or combination of failures.
In other news, I’m looking forward to Uber Eats saying “Jesus is on a jetpack” instead of his usual bicycle.
> The good news is that general aviation flying is getting safer. Accidents through 2018 are essentially flat over the preceding four years: about 1200 accidents per year and 200 fatal accidents per year.
>The general problem with relying on automation is that software developers can't necessarily anticipate and code for every possible failure mode. Whereas an experienced human pilot at least has a chance to invent new solutions in real time based on knowledge of first principles. For a prime example look at how Chesley "Sully" Sullenberger intentionally disregarded parts
We were promised jetpacks and instead we got a military-industrial complex funded by the promise of jetpacks. Can't wait til these trickle down to police departments.
How long does that jetpack fly? I had read somewhere that those engines use up 1 gallon per minute of flying. That seems like a lot. Imagine having to fuel up after every 5 minutes of flying.
Yes, they can. It'll cost more than the C152 is worth, but Garmin has a "press button to land plane" emergency feature recently approved by the FAA. It'll even announce itself to the relevant air traffic control frequencies.
> Simultaneously, Autoland analyzes terrain, weather, and nearby airports to determine the optimal airport for landing given the specific aircraft's available fuel and performance characteristics, winds, runway length, and a host of other factors. As the airplane descends, Autoland controls speed, altitude and flight path and manages throttles, flaps, cabin pressurization, and more.
> Air traffic controllers are informed of the system's landing destination choice and clear traffic as they would with a human-flown emergency. On final approach, Autoland lowers the gear, flares the airplane, and brings it to a stop with automatic braking. Audible instructions for exiting the airplane then play.
Maybe even cooler, in the sense of being interesting, is a European startup which is in the process of getting the first neural net certified by the FAA.
Reminds me of the nuisance at San Francisco airport. Where a guy with a jet pack caused f16s to dispatch. They never caught him. Had they, I don't think their first question would be why are you driving a jet pack. But rather how does it work for so long.
You're probably referring to a series of incidents at LAX, not SFO. I don't think they ever determined if it was a person in a jetpack or an unmanned drone built to resemble same.
How on earth do you escape in a jetpack. You own a jetpack. First, I'm sure the list of jetpack owners is very short, and second, it's not exactly a subtle object to own. Assuming the person has burred it in their garage since...
The general problem with relying on automation is that software developers can't necessarily anticipate and code for every possible failure mode. Whereas an experienced human pilot at least has a chance to invent new solutions in real time based on knowledge of first principles. For a prime example look at how Chesley "Sully" Sullenberger intentionally disregarded parts of the official checklist during the US Airways Flight 1549 emergency because he recognized that the procedure hadn't been written with their situation in mind.