Hello everyone.
Our team is currently trying to use Rectangular Target Processing in our Vision Processing loop. Although we were able to successfully integrate it into our code, it seems like the vision processing is slowing down all of the other code significantly. For example, when the vision processing is enabled, our motor code does not run fast enough, causing the motor to “skip” when it is supposed to be running continuously.
Can you recommend any ways with which we could optimize the vision vi? We are really using the color tracking in combination with reflective tape in order to get our target. We are also relying on the distance calculation as well as the target array that is created when a target is detected.
You can control the camera frame rate and the camera resolution. Additionally, closing the robot’s movement using only a camera is somewhat tricky due to lag and possible camera shake. It may be that you are using the camera too much or asking it to do too much.
Greg McKaskle
Many teams are trying to do the vision processing in a separate computer and sending the results to the cRIO (just the numbers, not the image). It’s a complicated solution that takes a lot of knowledge, and I’m not sure about its legality. In fact I have a few questions about it as well - is it legal to do the vision processing in the Dashboard (I mean, it already has the camera image, and the computer doesn’t do much else) and then sending it to the cRIO through UDP as discussed here?
(I’ll go cross-post it in that thread too)