"Ron Wanttaja" wrote in message
...
1. CAN it be done?

Hmm...yeah, probably, but the complexity of software that can handle what we
consider to be routine decision-making would be phenomenally high. Picture
what goes into a current state-of-the-art FBW software (incredibly
expensive......figure $80M for basic inner-loop and autopilot through to
certification) and take it up 2-3 orders of magnitude. We're talking about
stretching the limits of current AI tech, plus testing.

2. WILL it be done?

Nope. Not any time soon, anyway. There just isn't a strong enough tech push.

3. SHOULD it be done?

Not for GA alone. There just isn't a strong enough need.

The tech that you'd need to make it work are the same as those needed to
make UAVs trustworthy enough for "hands-off" operation within federal
airspace. If it will ever happen, it will start with UAVs, cuz that's where
the money is being spent. As long as UAVs are only adhering to their current
conops, this tech will never see enough attention to emerge into the
mainstream. It will remain the province of robotic deep-space exploration.

Pete