you're totally correct that emergencies are where the real issues lie. That's why we've built in multiple layers of fault tolerance so that a generator failure or a flight computer failure doesn't immediately revert you to a direct law control scheme (I'm assuming you're familiar with Airbus control laws given that you're using those phrases)
On top of everything, small GA aircraft have the luxury of being to use a ballistic parachute (which we will have) to bring the entire airframe to the ground in the event of a complete system failure. Which is always better than just letting the airplane crash into the ground.
I agree that proficiency is a fundamental issue. We want to make it easy to be and stay proficient. If flying becomes part of the primary mission, you do it more often and as a result stay proficient. The most dangerous pilots are the ones who haven't flown in 3 years and jump into a plane for an IFR cross country to an airport they've never been to before. We don't want that to happen either.
If you load up on redundancy - well, electronics redundancy it seems, since the rest of the aircraft seems to be a Sling TSi - you increase cost and weight, decrease speed and useable load, and so on, and now your aircraft costs $800k-1m and $100-150/hr to operate. How many new clients that would not otherwise have gone into GA will you attract? Dual navigation is already commonly available. So what are you doing different that YC was interested in paying for?
As our system has been developed so far, it has negligible impact on the speed and usable load of a base TSi, costing $500k for everything.
You mentioned elsewhere that redundancy means this is unlikely, but is there ever a reversion to "direct law"? Does the plane still have manual trim/rudder/throttle to deal with this, and are the pilots expected to periodically practice using the controls in "direct law"?
Do you have a working plane available for a test flight? :)
We've flown early versions in an test aircraft that we have and are currently integrating the latest version of Airhart Assist (seen in the video) into the Airhart Sling prototype to be flying and giving demos by the end of the year.
Our hope is that GA pilots will no longer need to do things like train stall recovery. We are moving into an era of aviation where aircraft (not just airplanes) will be complicated enough that computers have to be in the loop to handle things like that, because it allows us to create more interesting aerodynamic aircraft that are more efficient and have better performance.
regarding something like AF447, the immediate answer is we have the ballistic parachute as a final backup in case the pilot is unable to land the plane, for whatever reason. Realistically, it would depend entirely on the exact situation our plane was in, what systems have failed vs which haven't, and the pilots actual skill level
- even with redundancy, there can be enough failures that the computer no longer has enough information to make the right decision, one thing humans seem to be good at.
- people who had been trained in stall recognition and recovery, but probably had not kept proficient at it (unlike pilot of slow airplanes like GA and glider pilots) failed to recognize it (only 1 of the 3 pilot did) and did not recover from it.I hate everything about this thought process. Every pilot should know how to handle stall recovery, for the same reason that every driver should know how to handle loss of traction whether they have traction control or not. Driving the skill bar to the bottom intentionally will just result in more and more people flying without a clue what they're doing, just like we have on the roads today.
Have y'all considered that a fully digitized, fly-by-wire aircraft, of the type that you're proposing to build, would also essentially be a ridiculously high-fidelity flight simulator whenever and wherever it is sitting still with the engine off?
This could be part of your pitch—the pilot could program in a flight plan and then fly it in simulation, practice all the radio work etc., using the actual physical controls they would use to do it for real, then go do it for real all in the same chair.
Especially when it comes to practicing radio calls, it's the repetition that breeds competence and therefore confidence.
So, pilots who aren't legal to fly?
As a software engineer and private pilot, I'll take mechanical controls connected to the flight surfaces and my competency as a pilot to risk my own life, rather than turn it over to other software engineers, hoping they get everything right, leaving me with no real connection to the flight control surfaces when the shit hits the fan.
I'd at least have to know what kind of engineering process and change management practices or functional safety procedures are being followed by the company developing this stuff before I'd even consider going for a ride in such a plane.
As other commenters have point out though, where this stuff falls short is ultimately still the human. Ok, great… your aircraft won’t stall in Normal Law. However you’ve now lost a generator and a whole FCC and you’re in direct law. The 400 hr pilot hasn’t actually flown a plane with direct input since their primary training 5 years ago. They also don’t remember what the different flight envelopes do and do not provide. Essentially the system is more complex but normally it works so the complexity is hidden. They’re not equipped to handle flying the airplane anymore.
This is where GA really ultimately falls short IMHO, proficiency. Airlines are the safest they’ve ever been because the pilots make an entire career out of being prepared for every contingency. People using airplanes as a personal travel tool can be trained and proficient to the same degree but often they are not because flying an aircraft is ancillary to their primary mission.