buijs512 at planet.nl
Wed Jun 20 08:01:41 MST 2007
Antoine Beyeler wrote:
> Jon: no offence, but I'm not sure that's what the OP was looking for.
> I've not a very long experience in sensor modelling, but I know of
> people achieving reasonable success at implementing infra-red
> proximity sensors using a small number of rays (less than 5-10).
> There is in general a dependance on the material on which the IR is
> reflected, though.
> For infra-red triangulation-based distance sensor (e.g. the ones from
> sharp), a single ray should be sufficient alongside a noise model and
> a non-linear output mapping.
> For sonar it may be slightly more complicated since the aperture
> angle is usually larger and the output is usually proportional to the
> smallest distance within the aperture. Both of these issues call for
> a significantly larger number of rays. If you have 3D rendering built-
> in your software, you may consider modelling the sonar using the z-
> buffer of a camera with similar aperture angle, basically outputting
> the smallest distance from the z-buffer. If you want to go this way,
> I can provide some code example (OT to this list though).
Not sure what the exact intention is, but "cone tracing" may be of interest.
You can probably even implement it in ODE with the cone primitive but probably won't be fast.
More information about the ODE