Great idea, it seems you are going in a good direction! However, I do think asking users to keep in mind 4 (not that intuitive) gestures with 2 variations each (leading to a mental load of 8 actions) is too much... I think we could mix a couple of big touchable areas there, more graphic and distinguishable actions. Something easily distinguishable from a driver's glimpse.

It's a great idea--UI relative to the pointer rather than absolutely positioned in a frame.

Now I want to see how this is combined with the DISPLAY of information. What happens when you lift your fingers?

this is pretty rad.

it seems to require near-perfect knowledge of all options and ranges before becoming truly usable (something a knob with ticks communicates as part of it's form).

it also seems to penalize exploration (which would be required to mitigate the above), unless the adjustments require a commitment step (volume doesn't actually change until touch-out).

neat concept though.

"pinch three times with five fingers then scroll down 1/5 of the screen to turn on the rear defroster."

feels so great to see the right amount of design combined with refreshingly thoughtful interaction