ERO - Evolution of Robotical Organisms > Welcome
 
ERO - menu
  welcome 
  news 
  online demo 
  gallery 
  publications 
  contact 
NEWS

Sep 29, 2011
ERO presented at "One Day With Informatics" (details in Czech here)


Jan 20, 2011
Book chapter about novelty search in ERO accepted to Advances in Robotics and Virtual Reality (download here).


Dec 1, 2010
Paper on novelty search using ERO presented at ISDA'10, Cairo (download here).


Nov 3, 2010
Paper using ERO presented at CJS'10, Japan (download paper here).


Older news >>

 

ERO - Evolution of Robotical Organisms

ERO is a framework for evolutionary experiments, primarily used for experiments with 3D virtual creatures.


Features of the framework include:

  • distributed computation engine for computing CPU-intensive tasks

  • interactive real-time animation of evolved creatures (using ODE [ www.ode.org ] for rigid body dynamics simulation)

  • user-friendly editor and viewer of creature genotypes

  • advanced statistics of evolutionary process

  • browsable history of previous generations available during the evolution run

Current work

Currently, work on ERO continues at Charles University Computer Centre. Project now focuses on construction of evolved robots (click here for some pictures of preliminary hardware components) and large-scale evolutionary experiments (using on the order of hundreds of CPUs of Czech academic computer grid). People currently involved in the project are RNDr. František Mráz CSc., Mgr. Jindřich Stejskal, Mgr. Peter Krčah, Mgr. Daniel Toropila and Jan Koubek.


History

ERO was originally developed as a software project at Faculty of Mathematics and Physics, Charles University in Prague by a group of students (Peter Krčah, Daniel Toropila, Miroslav Blaško, Alexander Kutka and Rudolf Vlk). It has been inspired by virtual creatures developed by Karl Sims. The work on the project started in September 2004 and the project was sucessfully defended on June 23, 2006.


After that, the development of ERO continued in the thesis of Peter Krčah. Thesis introduced an application of one of the most successful algorithms for evolution of neural networks - NEAT - to the evolution of virtual creatures (full text of the thesis is available in the publications section). Thesis was successfully defended on September 19, 2007.


During its lifetime, ERO was successfully presented for various audiences (see news for details).

 
About      |       Contact       
ERO - Evolution of Robotical Organisms, robots, organisms, evolution, algorithms, computation, distributed computation, artificial intelligence, artificial life