Towards a Boolean network-based Computational Model for Cell Differentiation and its applications to Robotics

Living organisms are the ultimate product of a series of complex processes that take place within—and among—biological cells. Most of these processes, such as cell differentiation, are currently poorly understood. Cell differentiation is the process by which cells progressively specialise. Being a fundamental process within cells, its dysregulations have dramatic implications in biological organisms ranging from developmental issues to cancer formation. The thesis objective is to contribute to the progress in the understanding of cell differentiation and explore the applications of its properties for designing artificial systems. The proposed approach, which relies on Boolean networks based modelling and on the theory of dynamical systems, aims at investigating the general mechanisms underlying cell differentiation. The results obtained contribute to taking a further step towards the formulation of a general theoretical framework—so far missing—for cellular differentiation. We conducted an in-depth analysis of the impact of self-loops in random Boolean networks ensembles. We proposed a new model of differentiation driven by a simplified bio-inspired methylation mechanism in Boolean models of genetic regulatory networks. On the artificial side, by introducing the conceptual metaphor of the “attractor landscape” and related proofs of concept that support its potential, we paved the way for a new research direction in robotics called behavioural differentiation robotics: a branch of robotics dealing with the designing of robots capable of expressing different behaviours in a way similar to that of biological cells that undergo differentiation. The implications of the results achieved may have beneficial effects on medical research. Indeed, the proposed approach can foster new questions, experiments and in turn, models that hopefully in the next future will take us to cure differentiation-related diseases such as cancer. Our work may also contribute to address questions concerning the evolution of complex behaviours and to help design robust and adaptive robots.