Artificial Embryogeny and Grid Computing

In Artificial Life, the production of new artificial creatures always needs more and more computation power. Whereas artificial morphogenesis methods construct complete creatures using blocks, artificial embryogeny develops smaller creatures starting from a unique cell. To obtain a complete creature, organized in tissues and organs, we propose a developmental model in which cells are coded as threads. This massive parallel architecture allows the simulation of an organism development on multi-core or multi-processor machines. In most cases, evolutionary algorithms and especially genetic algorithms are used to create our creatures. Their algorithms take a lot of computation time to find an environment-adapted creature. In order to reduce the computation time, genetic algorithms have already been parallelized, but, in most cases, using a supercomputer. This solution is very expensive and not easily scalable. In this report, we first present our model of artificial embryogeny, Cell2Organ. Then, we propose an implementation of genetic algorithms for artificial embryogeny using a computational grid and ProActive middleware.