Genetic encoding of robot metamorphosis: How to evolve a glider with a genetic regulatory network

Abstract

In REPLICATOR [2] powerful reconfigurable robots are designed and constructed. Reconfigurable robots can dock together and form robot organisms. Robot organisms have the ability to morph from snakes into spiders, chairs, swarms, wheels. The problem we are facing is: How to evolve self-organized robot metamorphosis? The metamorphosis graph A={S,T} is a tuple of S, the set of all possible robot module configurations, and T the set of all transitions between those configurations. A configuration s S, s={R,D} consists out of R robots with D connections. If D=Ø, s denotes a swarm. A fitness function f for reciprocal metamorphosis defines a maximum for a specific cycle in A. A metamorphic fitness function is used that defines maximum fitness for a dynamic body form called a glider: a snake growing its head, and losing its tail a g A. The evolutionary search process needs to find a self-organized solution for a glider in A. © 2010 Springer-Verlag Berlin Heidelberg.