Quote:
|
Originally Posted by kmazh
I had an impression (by reading the setup guide) that you have to "teach" the camera by giving it a target image. Looks like GUI can convert the target image into set of parameters to be loaded to the camera. Is it correct understanding? My idea was to track the ball when you shoot it and track the goal when you drive to it.
Konstantin
|
First, you kind-of "teach" the camera what the thing you want to track looks like. In this case, the green light box. You do this by downloading new parameters into it.
Second, you should have no problems tracking the green light box, but I very much doubt you will be able to track the ball in flight. The issue is that the color is not constant - not only do the balls come in different colors, they are half black and half red (for example) and so the camera doesn't know what to track - red or black.
Your understanding of the camera function is reasonably correct. You teach it a color to track, and it returns not an image but a set of coordinates, telling you where the target is relative to (something)-- 26 degrees up and 13 degrees to the left, for example.
That's a bit of a simplification, but not much. Read the CMUcam book from usfirst.org
Don