more about evolving bodies than just brains, but both are interesting. Areas such as simulated physics, modelling of machines, building from design, etc are all appropriate, moreso the real world errors that creep in.

• http://asl.epfl.ch/research/projects/EvolutionaryRobotics/

Old-school

• Karl Sims http://web.genarts.com/karl/evolved-virtual-creatures.html
• Tom Ray http://www.his.atr.co.jp/~ray/pubs/alife7a/index.html

Current

• Framsticks http://www.frams.poznan.pl/
• Spiderland/Breve http://www.spiderland.org/breve/
• Hod Lipson, the daddy of realworld building from evolutionary stuff http://www.mae.cornell.edu/lipson/
• DEMO lab http://www.demo.cs.brandeis.edu/
• Greg Hornby http://www.demo.cs.brandeis.edu/pr/evo_design/evo_design.html
• Realworld swarming doesn't work right: Nik has these links somewhere

reading

• http://www-106.ibm.com/developerworks/wireless/library/wi-robotics.html?ca=dgr-lnxw01WirelessRobots

A multidisciplinary research group at the University of Calabria (Italy) is working on evolutionary robotics in which neural network and genetic algorithms are applied to select the fittest behaviour. Robots may change during lifetime and adapt to their (varying) environment. Collaboration with industrial research centres and universities is sought for a joint development into specific application fields.

Evolutionary Robotics is a technique for the automatic creation of autonomous robots that is inspired by the Darwinian principle of selective reproduction of the fittest. The basic idea behind evolutionary robotics is the following:

• An initial population of different artificial genotypes, each encoding the control system and sometimes the morphology of a robot, are randomly created and put in the environment.
• Each robot is then let free to act (move, look around, manipulate) according to a genetically specified controller while its performance on a given task is automatically evaluated.
• The fittest robots are allowed to reproduce by generating copies of their genotypes with the addition of changes introduced by some genetic operators (e.g., mutations, crossover, duplication) .
• This process is repeated for a number of generations until an individual is born which satisfies the performance criterion (fitness function) set by the experimenter.

The technique is based on the selection of neural networks by means of genetic algorithms, on the basis of their competence to achieve a given objective. A behaviour can be first selected in a software simulated environment and then the “emerging” behaviour can be applied to a physical robot and tested on a real ground.

Robots might also change during lifetime and therefore might adapt to their environment. Artificial evolution might lead to the development of simple and effective solutions also in the case of problems that have to be solved in varying environmental conditions.

Main Advantages:

• Robots can operate in new, modified and evolving environments.
• Robots and their algorithms can be developed in controlled conditions and then tested in real environment.

Innovative Aspects: It is a new approach that looks at robots as autonomous artificial organisms that develop their own skills in close interaction with the environment without human intervention.

Contact Details

• Contact Name : ROSSI, Giuseppe
• Contact Organisation : Consorzio Catania Ricerche
• Email: irc.calabria@tiscali.it
• URL: http://www.unict.it/ccr/eng/home.html