Applying biological concepts to create new models in the computational field is not a revolutionary idea: science has already been the basis for the famous artificial neuron models, the genetic algorithms, etc. The cells of a biological organism are able to compose very complex structures from a unique cell, the zygote, with no need for centralized control (Watson J.D. & Crick F. H. 1953). The cells can perform such process thanks to the existence of a general plan, encoded in the DNA for the development and functioning of the system. Another interesting characteristic of natural cells is that they form systems that are tolerant to partial failures: small errors do not induce a global collapse of the system. Finally, the tissues that are composed by biological cells present parallel information processing for the coordination of tissue functioning in each and every cell that composes this tissue. All the above characteristics are very interesting from a computational viewpoint. This paper presents the development of a model that tries to emulate the biological cells and to take advantage of some of their characteristics by trying to adapt them to artificial cells. The model is based on a set of techniques known as Artificial Embryology (Stanley K. & Miikkulainen R. 2003) or Embryology Computation (Kumar S. & Bentley P.J 2003).
The Evoluationary Computation (EC) field has given rise to a set of models that are grouped under the name of Artificial Embryology (AE), first introduced by Stanley and Miikkulainnen (Stanley K. & Miikkulainen R. 2003). This group refers to all the models that try to apply certain characteristics of biological embryonic cells to computer problem solving, i.c. self-organisation, failure tolerance, and parallel information processing.
The work on AE has two points of view. On the one hand can be found the grammatical models based on L-systems (Lindenmayer A. 1968) which do a top-down approach to the problem. On the other hand can be found the chemical models based on the Turing’s ideas (Turing A. 1952) which do a down-top approach.
On the last one, the starting point of this field can be found in the modelling of gene regulatory networks, performed by Kauffmann in 1969 (Kauffman S.A. 1969). After that, several works were carried out on subjects such as the complex behaviour generated by the fact that the differential expression of certain genes has a cascade influence on the expressions of others (Mjolsness E., Sharp D.H., & Reinitz J. 1995).
The work performed by the scientific community can be divided into two main branches. The more theoretical branch uses the emulation of cell capabilities such as cellular differentiation and metabolism (Kitano H. 1994; Kaneko K. 2006) to create a model that functions as a natural cell. The purpose of this work is to do an in-depth study of the biological model.
The more practical branch mainly focuses on the development of a cell inspired-model that might be applicable to other problems (Bentley, P.J., Kumar, S. 1999; Kumar, S. 2004). According to this model, every cell would not only have genetic information that encodes the general performance of the system, it would also act as a processor that communicates with the other cells. This model is mainly applied to the solution of simple 3D spatial problems, robot control, generative encoding for the construction of artificial organisms in simulated physical environments and real robots, or to the development of the evolutionary design of hardware and circuits (Endo K., Maeno T. & Kitano H 2003; Tufte G. & Haddow P. C. 2005).