We created a new tool for 3D architectural design. Essentially, we hooked GEVA (the NCRA's grammatical evolution software) up to Blender 3D, a 3D modelling package. Using interactive evolution, the user could explore families of architectural designs. We wrote a library of different grammars, each expressing a different language of possible designs.
I also came up with a new, continuous-valued, fixed-length encoding for grammatical evolution itself. It's intended to be of benefit to numerical approaches such as interpolating crossover and PSO-like algorithms in the context of GE.
I collaborated with Edgar Galvan-Lopez, Michael O'Neill and Anthony Brabazon on a strand of research concerning representations and measures of problem difficulty in standard GP. Of particular interest was the idea of the locality of the mapping from genotype to phenotype and thence to fitness.
As side projects, I did some work on evolutionary art and music.
I supervised students working in evolutionary music, and in computational linguistics.