A group of researchers at Michigan State University
is investigating how to harness the power of
evolutionary computation
to help design robust, cooperative distributed systems.
Many of our early studies employed
digital evolution,
whereby a population of self-replicating computer programs evolves
through natural selection, enables system designers to explore an
enormous solution space for complex problems, often yielding strikingly
clever results.
In particular, the Avida platform
provides a digital Petri dish for discovering new computational behavior.
Evolved algorithms (sequences of instructions
comprising the genomes of digital organisms) can simply be recompiled to
execute on collections of distributed devices, including sensor nodes and
micro-robots.
The Thinktank initiative
involves collaboration among researchers in computer science,
microbiology, electrical engineering, and mechanical engineering.
These collaborations were part of a multidisciplinary effort that
led to the establishment of
BEACON, an NSF Science and Technology Center for the Study of
Evolution in Action.
Our recent studies have applied several additional forms of evolutionary
computation to the development of dynamically-adaptive systems.
Methods include genetic algorithms,
genetic programming, neuro-evolution (evolving artificial neural networks).
Experiments are conducted in Evolution Park, an evolutionary robotics
habitat
for crawling, swimming, and flying creatures.
The results of
our investigations, to date, are described in several
publications.
Early work in this area is summarized in a
January 2008 IEEE Computer paper, Harnessing Digital
Evolution.
