Information technology has only been around for about fifty years. Although the beginnings of automatic calculation date from as early as the 17th century (W. Schickard built the first mechanical calculator in 1623), it took the invention of the transistor by W. Shockley, J. Bardeen and W. Brattain in 1947 to catapult calculators out of the laboratory and produce the omnipresence of information and communication systems in today's world. Computers not only boast very high performance, capable of carrying out billions of operations per second, they are taking over our world, working their way into every last corner of our environment. Microprocessors are in everything, from the quartz watch to the PC via the mobile phone, the television and the credit card. Their continuing spread is very probable, and they will even be able to get into our clothes and newspapers. The incessant search for increasingly powerful, robust and intelligent systems is not only based on the improvement of technologies for the manufacture of electronic chips, but also on finding new computer architectures. One important source of inspiration for the research of new architectures is the biological world. Nature is fascinating for an engineer: what could be more robust, intelligent and able to adapt and evolve than a living organism? Out of a simple cell, equipped with its own blueprint in the form of DNA, develops a complete multi-cellular organism. The characteristics of the natural world have often been studied and imitated in the design of adaptive, robust and fault-tolerant artificial systems. The POE model resumes the three major sources of bio-inspiration: the evolution of species (P: phylogeny), the development of a multi-cellular organism by division and differentiation (O: ontogeny) and learning by interaction with the environment (E: epigenesis). This thesis aims to contribute to the ontogenetic branch of the POE model, through the study of three completely original cellular machines for which the basic element respects the six following characteristics: it is (1) reconfigurable, (2) of minimal complexity, (3) present in large numbers, (4) interconnected locally with its neighboring elements, (5) equipped with a display capacity and (6) with sensor allowing minimal interaction. Our first realization, the BioWall, is made up of a surface of 4,000 basic elements or molecules, capable of creating all cellular systems with a maximum of 160 × 25 elements. The second realization, the BioCube, transposes the two-dimensional architecture of the BioWall into a two-dimensional space, limited to 4 × 4 × 4 = 64 basic elements or spheres. It prefigures a three-dimensional computer built using nanotechnologies. The third machine, named BioTissue, uses the same hypothesis as the BioWall while pushing its performance to the limits of current technical possibilities and offering the benefits of an autonomous system. The convergence of these three realizations, studied in the context of emerging technologies, has allowed us to propose and define the computer architecture of the future: the electronic paper.