Steve,

If moving the sensor a lot within its NxN field is ok, what's wrong with doing a non-tangent X and Y slice to compute the center?

Think of it as a geometric problem of drawing a vertical line and a horizontal line across a circle. Since the lines go void to void across the circle, each line's midpoint reveals a coordinate of the circle's centerpoint.

While this requires moving the sensor more than I would consider "elegant", it seems fairly easy.

(1) move the sensor into the beam until a threshold intensity is reached to avoid cutting a near tangent slice of the beam. This is the starting point. Thresholds can be taken from this discovery.

(2) Move the sensor along the X axis until it reaches the void region on one side. Now slice in the reverse direction until your X axis slice reaches the opposite void. The midpoint of the slice is the position X coordinate of the beam center due to the beam's symmetric shape.

(2a) Datapoints along the path can either be kept or merely look for intensity above a very low threshold and and bit map the illuminated width of the slice. That makes it a simple matter of mapping the centerpoint of the above-threshold run from void to void.

(3) Repeat for a Y axis slice. Midpoint is the position Y coordinate of the beam center.

(4) Move to the XY coordinate from the two slice midpoints.

Its a quick an easy solution. I'd prefer to do a good self centering algorthm taking advantage of its gaussian distribution... but that takes project time if its available.

-Jay C. Box (still)