Evolving embodied genetic regulatory network-driven control systems

Abstract

We demonstrate the evolution of simple embodied Genetic Regulatory Networks (GRNs) as real-time control systems for robotic and software-based embodied Artificial Organisms, and present results from two experimental test-beds: homeostatic temperature regulation in an abstract software environment, and phototactic robot behaviour maximising exposure to light. The GRN controllers are continually coupled to the organisms' environments throughout their lifetimes, and constitute the primary basis for the organisms' behaviour from moment to moment. The environment in which the organisms are embodied is shown to play a significant role in the dynamics of the GRNs, and the behaviour of the organisms.