It exports to wavefront .obj files, which are generally a 3d industry standard file. I have 5 full 3d modelling packages, and 3 ray - tracer packages, and 1 terrain modelling package, these all accept .obj files.
Thinking along these lines, the general concept seems an excellent way to attempt adding object recognition into a robot, by using the camera to first take the outline of the object, quickly laser scan it and you could end up with 3d object data rather than just a 2d object. Its a bit ambitious for microcontrollers, and the laser scanner would have to be mounted in a way to have full pan, tilt, and translate. distance recognition without the special background would be difficult, but easier with 2 camera's. I think i'll leave it on the back burner for now.....
Possibly there is some other way without a laser, again 2 cameras would be a way forward.