Evolving Insect Locomotion Using Non-uniform Cellular Automata

Abstract

We present a model for the evolution of locomotion behavior in a simulated insect. In our model, locomotion is defined over a discrete state space using non-uniform cellular automata. We use a genetic algorithm to evolve the dynamical law of the system. The model was capable of evolving several commonly observed gaits of insects. In addition, the evolutionary process yielded periodic at-tractors which are invariant from the initial conditions.