more on Nature 31 aug 2000
http://www.nature.com/nature/fow/
then click "Robotic Vision"....
Robotic Vision
Two papers in this week's issue of Nature push forward the task of creating
robots which mimic the way living creatures evolve and behave. Lipson and
Pollack report a system which designs, 'evolves' and manufactures simple
robots with almost no human intervention. A computer simulation using
genetic algorithms which mutate a simple body plan was used to design robots
capable of horizontal motion. The computer system could then test these
simulations in the physical world by building the robots using a
rapid-prototyping technology. In an accompanying News and Views article
Rodney Brooks explains how this research contributes to the development of
self-evolving machines. In a separate study Krieger et al show that robots
programmed to co-operate in an ant-like manner are able to complete a
foraging task in a more efficient manner than single robots. Their Khepera
robots were found to recruit assistance from other robots, in the same way
ants co-operate to gather food, when they came across a resource cluster.
Both sets of robots can be seen in action by following the links to the
supplementary information provided by the authors.
Ant-like task allocation and recruitment in cooperative robots
MICHAEL J. B. KRIEGER, JEAN-BERNARD BILLETER & LAURENT KELLER
Automatic design and manufacture of robotic lifeforms
HOD LIPSON AND JORDAN B. POLLACK
Artificial life: From robot dreams to reality
RODNEY BROOKS
Can evolutionary principles be used to automate machine design? A new study
evolves robot body parts and nervous systems inside a computer, and then
physically builds them, so giving birth to real machines created by virtual
evolution.
Genome complexity, robustness and genetic interactions in digital organisms
RICHARD E. LENSKI, CHARLES OFRIA, TRAVIS C. COLLIER & CHRISTOPH ADAMI
Evolution: Creatures from another world
INMAN HARVEY
Digital organisms are computer programs that are designed to mimic
evolutionary processes by mutating and reproducing. Their use is
controversial, but they might allow scientists to ask questions about
evolution that cannot be answered easily by studying real organisms. Such
organisms have been used to study how multiple mutations in the same
organism interact, with combined effects that can be greater or less than
the sum of their parts.
Inspiration for optimization from social insect behaviour
E. BONABEAU, M. DORIGO & G. THERAULAZ
Dynamical & Evolutionary Machine Organization
http://www.demo.cs.brandeis.edu/
----- Original Message -----
> This story was on the front page of today's New York Times.
> Bill Joy is quoted, he's very very unhappy. I think the reporter
> who wrote the story is unhappy too, the doomsday tone in the
> article is almost as strong as when the first report of Dolly was
> splashed on the front page of that same newspaper.
>
> John K Clark jonkc@att.net
>
> http://www.nytimes.com/library/tech/00/08/biztech/articles/31robot.html
This archive was generated by hypermail 2b29 : Mon Oct 02 2000 - 17:36:52 MDT