Epigenetic adaptation in action selection environments with temporal dynamics

Abstract

To operate in dynamic environments robots must be able to adapt their behaviour to meet the challenges that these pose while being constrained by their physical and computational limitation. In this paper we continue our study into using biologically inspired epigenetic adaptation through hormone modulation as a way to accommodate the needed flexibility in robots’ behaviour, focusing on problems of temporal dynamics. We have specifically framed our study in three variants of dynamic three-resource action selection environment. The challenges posed by these environments include: moving resources, temporal and increasing unavailability of resources, and cyclic changes in type and availability of resources related to cyclic environmental changes.