I used a Monte Carlo method of finding probabilities of collisions between a randomly
tossed circular block and non-overlapping randomly placed circular targets on a
3200 x 3200 grid:
http://home.ptd.net/~artnpeg/targets.htm
The page has further descriptions and info. In order to find a mean p, I've found that 100 runs averages "home in" fairly well using the default parameters. 100 runs takes about a minute on my computer.

The page details how to set Firefox to allow sufficient javascript run time. With Google Chrome, it's just a matter of clicking "Wait" once while the script executes. With Opera, I didn't experience any warning screens.

I got interested in this kind of problem because of engineering and scientific
applications.

Art