Files

Abstract

Nature is an incredibly complex system resulting from the interactions of many different interdependent species, composed of many individual organisms, which are in turn composed of an extremely large number of cells. In multi-cellular organisms, the parallel operation of millions or trillions of cells is an astounding example of massively parallel processing which results in the emergence of extremely complex global features such as robustness (organisms can tolerate a substantial amount of damage), cicatrization (organisms can self-repair), regeneration (some organisms can regrow certain parts of their bodies), adaptation (through death and replacement of their cells, organisms can cope with a changing environment) and development (organisms grow during their life). Because it underlies these interesting behaviors, multi-cellular organization is a very tempting source of inspiration for the design of artificial systems and a growing number of engineers are beginning to look at it to find inspiration on how to adapt its basic paradigms to electronic computing machines. The POE model defines the three major axes of this bio-inspiration: Phylogenesis (the evolution of the species), Ontogenesis (the development of a complete organism from a single cell as directed by its genetic code), and Epigenesis (the learning processes by interaction with the environment). This thesis aims to contribute to the ontogenetic branch of the POE model through the proposition of different hardware implementations of self-replication strategies that can be integrated in the design of new programmable devices, which in turn could enable structures to replicate within these electronic substrates. To validate the different approaches of self-replication proposed, i.e. construction-based, self-inspection based and self-repairing, each algorithm is applied to a multi-processor system based on the MOVE architecture and implemented in a modified version of the POEtic tissue, a custom programmable device. The different strategies which are proposed could then provide new avenues of exploration for the implementation of bio-inspired systems, for conventional computing systems and, perhaps, even for next-generation electronics based on nanotechnology.

Details

Actions