I'm a member of FRC Team 6936. We have installed and tested the Raspberry Pi FRC Console for driver vision, and we're looking to add functionality using computer vision to detect the vision targets that are placed on all field elements. I know nothing about openCV, but one of the mentors pointed me to it as a possibility. Can OpenCV be used for real-time image processing, e.g. to point out to an operator when the robot is correctly aligned for a certain task? Can OpenCV work in parallel with the vision console or would that have to be run on a separate Pi? (we have 2)
Our goal is to have the actual processing be done on the Raspberry Pi, so as to reduce the amount of information sent over the wireless network and adhere to the 4 mbps network limitation. We already have two cameras running for driver vision, which consume about 2 mbps together, and we need some left over for robot control and diagnostics.
With this in mind, our original plan was to use a PixyCam to detect the retroreflective vision targets. However, PixyCam can't be used with the Pi (as far as I know).
Can the same stream of video be used for streaming to the operator and for vision processing?
Anyway, thanks.