Vision with GRIP

Hello everybody,
We are team RoboActive #2096 and we are a new team.
We are trying to implement a vision processing with GRIP.
I followed by this tutorial: http://wpilib.screenstepslive.com/s/4485/m/50711

and it explains how to do the process very nice, but the camera connected straight to the computer and not to the roborio.
My question is, how I getting the video from the camera while it connected to the roborio to do the process in GRIP?

See this tutorial to deploy GRIP to the roboRIO, and this tutorial to run GRIP from robot code.