Evolving virtual creatures revisited

  • Authors:
  • Peter Krčah

  • Affiliations:
  • Charles University, Prague, Czech Republic

  • Venue:
  • Proceedings of the 9th annual conference on Genetic and evolutionary computation
  • Year:
  • 2007
  • Evolving virtual creatures

    SIGGRAPH '94 Proceedings of the 21st annual conference on Computer graphics and interactive techniques

Quantified Score

Hi-index 0.00

Visualization

Abstract

Thirteen years have passed since Karl Sims published his work onevolving virtual creatures. Since then,several novel approaches toneural network evolution and genetic algorithms have been proposed.The aim of our work is to apply recent results in these areas tothe virtual creatures proposed by Karl Sims, leading to creaturescapable of solving more complex tasks. This paper presents oursuccess in reaching the first milestone -a new and completeimplementation of the original virtual creatures. All morphologicaland control properties of the original creatures were implemented.Laws of physics are simulated using ODE library. Distributedcomputation is used for CPU-intensive tasks, such as fitnessevaluation.Experiments have shown that our system is capable ofevolving both morphology and control of the creatures resulting ina variety of non-trivial swimming and walking strategies.