Research

    The focus of my research is emergent behaviour in cellular automata and what useful computation this can achieve. The area I am interested in applying this is to guide a mobile robot with vision.

    Why cellular automata? They are simple mathematical systems whose discrete state components are distributed regularly over space and interact locally. Because they capture the essence of influence spreading through neighbouring parts they can be used to model many natural systems and we hope to use them in engineering. Why emergent behaviour? CA demonstrates the whole is not the sum of its parts and global behaviour emerges from the local interactions of the components. This fact points to a basic lesson in science that the whole picture has to be taken account of not just reducing a system to its parts.

    To explore the emergent behaviour a mathematical filter for CA has been developed to remove repetitive structure, although one that can contain random elements, to reveal hidden coherent structure. This has been extended from the 1D to 2D case. A method for designing CA with a chosen repetitive structure is also being developed for the 2D case.

    In others work interacting coherent structures in 1D CA, found using artificial evolution, have provided a means to classify whether the initial condition contain a majority of black or white cells. Our aim is to tentatively extend such emergent computation ability to visual processing tasks in robotics.