[ODE] Hovercraft physics inside a tunnel

[Jon Watte]

> - The glider starts to "roll up" the walls like a pendulum, gaining more and
> more sideways speed when just left alone. I currently apply a gravity of -9
> and no significant linear dampening. Whats the cause for this? Would
> dampening help?

Linear and, more importantly, angular dampening is the quickest way to 
get a well-behaved anti-gravity glider. I've done that in the past. I 
also found that I want to scale down the force applied by each ray by 
how much above the center of the glider the ray position is; this will 
work to keep the glider level.

Also, I find that it works better if I take into account the angular 
velocity of the glider and the time step size. I apply the current 
angular velocity as a forward prediction for a few steps (like 3), and 
see what the rotation will be then, and I compare that to what I want 
the rotation to be (canceling out rotation around "up"); then I apply 
forces according to the difference between the two. This is a simple 
error-feedback controller with look-forward, which works very well.

Cheers,

			/ h+