How 'electronic tissue' will transform robotics

HAL - the thinking, talking computer that ran the spaceship in the film 2001: A Space Odyssey - was a brilliant concept but it now pales into insignificance compared to what is just around the computational corner.

Future generations of computers and robots will be based on an electronic tissue that mimics plant and animal tissue in its ability to grow and repair itself.

This means the machines will be capable of evolution, growth, self-repair, self-replication and learning.

Hard at work on creating this tissue that will be the essential substrate of these truly extraordinary creations is an international team coordinated by British electronics experts. Put quite simply, their goal is the development of a novel digital electronic circuit that combines the characteristics of both a flexible computational substrate and an artificial tissue.

This extraordinary project is aptly named POEtic, this being an acronym for the final aim of the circuit integrating the three biological models of self-organisation:

• Phylogenesis (P), the history of the evolution of the species;
• Ontogenesis (O), the development of an individual as directed by their genetic code;
• Epigenesis (E), the development of an individual through learning processes (nervous system, immune system) influenced both by their genetic code (the innate) and by the environment (the acquired).

These three models share a common basis; a one-dimensional description of the organism, the genome.

Ultimately the tissue will be the essential substrate for the creation of POE-based computers.

"While each of these models, taken separately, has to a greater or lesser extent been used as a source of inspiration for the development of computing machines, their amalgamation into a single artefact is a challenge yet to be met," said Prof Andy Tyrrel, of the Department of Electronics, at the University of York, England.

"It is our hypothesis that such a machine could be created with the support of a flexible computational substrate. Thus the goal of the POEtic project is the development of a flexible computational substrate inspired by the evolutionary, developmental and learning phases in biological systems."

The POEtic tissue will be a cellular surface composed of a variable number of elements or cells.

Each cell will contain the entire description, or genome, for the whole tissue and will have the ability to communicate with the environment, through sensors and actuators, and with neighbouring cells, through bi-directional channels.

All this means that it will be able to carry out a task.

In fact, during the project the tissue will be proved by carrying out various functions and applications that will benefit from its lifelike qualities and so be used to implement different multi-cellular organisms capable of interacting with and adapting to their own environment.

Each cell of the tissue will have the same basic structure but, uniquely, will be able to acquire different functionalities, similar to totipotent cells in living organisms. This flexibility will be provided by an organisation that is composed of three layers: a genotype plane, a configuration plane, and a phenotype plane.

The genotype plane of each cell will contain a full description of the organism in the form of a digital genome.

The configuration plane will transform the genome into a configuration string directly controlling the processing unit of the phenotype plane.

Through this cellular process the tissue will be organised into a massively parallel multi-cellular electronic structure.

Within such structure, groups of cells will be able to cooperate to realise a given task, giving rise to substructures not unlike organs in living beings.

The decomposition of the POEtic tissue into cells will allow the formation of the organism through a growth process inspired on the embryology mechanisms of cellular division and differentiation, and of a self-repair process through self-replication of cells on the tissue.

These processes will represent the ontogenetic model (O). The layered organisation of each cell will support different types of adaptive mechanisms. The genotype place will support the execution of genetic operations, such as cross-over and mutation to evolve the tissue functionality (phylogenetic inspiration: P model).

The configuration plane will allow the implementation of different types of learning mechanisms on the phenotype (epigenetic inspiration: E model).

For example, some cells could be configured as artificial neurones and consequently that part of the tissue could act as a neural network on which neural learning algorithms can be applied.

Other cells could take different functions, for example exploiting their input/output capabilities to provide an interface between the environment and the neural network, or monitoring of metabolic functions.

Another adaptation mechanism could be inspired by the animal immune system to improve the self-repair capabilities of the tissue.

The final tissue evolved from this will be the essential substrate for the creation of POE-based machines, capable of evolution (P model), of growth, self-repair and self-replication (O model), and of learning (E model).

This substrate will consist of one or more VLSI (very large scale integration) chips with high levels of connectivity between cells in the chip(s) and also high levels of interaction with the environment, for example input and output sensors and actuators.

Also taking part in the project are the universities of Glasgow, Scotland; Catalunya, Spain; and Lausanne, Switzerland.

Among the possible applications of POE-based machines are in the disparate fields of intelligent musical synthesisers; autonomous, intelligent robotics; and in long-term space missions where reliability, adaptability and autonomous behaviour are critical.

Prof Tyrrell hopes that a real demonstrator will be working by the end of the three-year project and it could well be possible that these ideas might be in everyday use in about five to 10 years.