Autonomous car: - drives pretty much on a plane - able to stop when confused - precise speed and direction control - stable (i.e. does not oscillate or slide) - uses the same environment (the actual road) for reference and "handling" (tire grip) - anyone with IQ > 70 can drive one

Autonomous sailplane: - flies in 3D space with pitch, roll, yaw - has both MAX (Vne) and MIN (stall) speed, can't stop and wait - can't even stop descending (!) - HEAVILY influenced by wind and thermal conditions - pretty unstable, oscillates when upset - can not use ground as visual reference reliably - flying one is no joke even for a human, requires serious training

Good luck with your full autopilot.

Autonomous car is harder than autonomous plaine. An autonomous plane can be flown with near-zero accidents by dGPS and traffic control alone. Try that with a car ;) . But take an autonomus plane, add an atmospheric model - you can have an autonomous sailplane.

It can be FLOWN with near zero-accidents by dGPS and control alone. That's steady/level flight, not counting issues with weather, launching, landing, and responding to in-flight emergencies. In a driving failure case, you crash or slow down, and you generally keep the accident constrained to the road. In a flying failure case, you've got a high-speed projectile moving in 3 directions.

Flown = Takeoff, flight, landing, emergencies handling.

What pilots usually do during emergency is to go through the checklist. There is not much room for improvisation.

Vehicles - it is a completely different level of problem. If you try to drive with dGPS alone, you will collide almost immediately. Even on the empty road, because dGPS is not good enough. And actually, it is almost AGI-complete problem, as you need to model and interact with other traffic participants.

> Good luck with your full autopilot.

While it doesn't completely negate your argument...

https://www.raspberrypi.org/blog/zero-hab-glider/

http://izzybrand.com/