Quote:
|
Originally Posted by Al Skierkiewicz
I am less concerned about a false trigger on the vision system from green anythings, (including humans) than I am about variations in lighting from venue to venue. When a material looks green to you it may not appear green to a camera depending on the color of the light. It may not even appear to a camera if the light is reflecting in the right way. I think that if you read the paper on visualization you might get a better handle on how this recognition works. I would think a robot would only get in trouble if it had a green stripe of approx the same height and width as the tetra and the same altitude and texture. What I am saying is test like crazy to be sure.
|
Excellent point.
It will be difficult to test like that when you're concerned about other robots, though.
Think about it--that robot followed Dave around, and it also followed the green tetra, if I'm not mistaken. Those were wildly different greens.
--Petey