Frontiers in Robotics and AI (Mar 2015)
Incremental Embodied Chaotic Exploration of Self Organised Motor Behaviours with Proprioceptor Adaptation
Abstract
This paper presents a general and fully dynamic embodied neural system, which incrementallyexplores and learns motor behaviours through an integrated combination of chaotic search andreflex learning. The former uses adaptive bifurcation to exploit the intrinsic chaotic dynamicsarising from neuro-body-environment interactions, while the latter is based around proprioceptoradaptation. The architecture developed here allows realtime goal-directed exploration andlearning of the possible motor patterns (e.g. for locomotion) of embodied systems of arbitrarymorphology. Examples of its successful application to a simple biomechanical model and asimulated 3D swimming robot are given. The tractability of the biomechanical systems allowsdetailed analysis of the overall dynamics of the search process. This analysis points out strongparallels with evolutionary search.
Keywords
