Camera driving means that the image processing looks for edges, trying to decide if there is an object to be passed... the ER1 software has the possibility to use 2 cameras, one for object recognition and one for driving. But if you use the API, you can swap cameras and only use one camera for one purpose at a time. For instance, look around to find the object you need to grab or go to, then change the function of the camera for driving towards the object, avoiding obstacles in the way. I just tilt down the camera to the proper angle for driving, because the camera has to be calibrated when the robot starts. Actually, it has to be done once, because it saves the configuration immage and it compares the frames it gets with the config frame.
