Yes, we have designed and built the aiming system, and should be testing it this week.
We have a rotating turret, to which the camera is attached. The camera is queried by the robot controller, and the turret is turned one way or the other to keep the camera aimed at the green light, irrespective to the orientation of the robot chassis.
For elevation control, we use a curved deflector that changes the direction of the ball from vertical to horizontal - or some angle in between. This is controlled by the threaded shaft in the KOP. A modified Globe motor (5 x normal speed by bypassing one layer of its planetary gear train) rotates the shaft to change the elevation. The degree of elevation is computed by reading the target Y axis info from the camera's vertical servo, which we retain, in combination with the measured target area, which gives us an approximation of the range to the target. We ditched the X servo, as the turret rotation serves this function now. We plan to empirically generate a lookup table to correlate the camera elevation and target size to the required elevation of the curved deflector.
We are hoping that this will give us automatic, on-the-fly targeting capability, with no operator interaction. Of course, theory and practice seldom have anything to do with each other.
We'll find out more this week, and will post videos if things work out.