Last year, I wrote about some code I had written to find maximally distributed sets of colours in perceptual colour space. I was having some problems with the code at the time, but since then I have got it working. I fixed it by only repelling each particle from the nearest one to it - then particles quickly settle into the points where they are equidistant from the nearest particles on each side.
Here are the colours it came up with in LUV space:
And here are the colours it came up with in LAB space:
I also did a variation where it just chooses a different hue for each colour, maximizing the saturation (so arranging colours in a ring, rather than throughout a volume) - this is the LAB version but the LUV version is very similar:
I want to do a little flash applet so you can see how the colour particles repel and rotate them around in 3D space - it's a very good way to visualize the perceptual colour solids.