2008
Pilat, Marcin; Jacob, Christian
Creature academy: a system for virtual creature evolution Proceedings Article
In: CEC 2008, IEEE Congress on Evolutionary Computation, pp. 3289-3297, IEEE Press, 2008.
Abstract | Links | BibTeX | Tags: artificial life, evolutionary design
@inproceedings{pilat2008wcci,
title = {Creature academy: a system for virtual creature evolution},
author = {Marcin Pilat and Christian Jacob},
doi = {10.1109/cec.2008.4631243},
year = {2008},
date = {2008-07-01},
urldate = {2008-07-01},
booktitle = {CEC 2008, IEEE Congress on Evolutionary Computation},
pages = {3289-3297},
publisher = {IEEE Press},
abstract = {In this paper, we present Creature Academy, a virtual laboratory that allows for the evolution of form and function within simulated physical 3D environments. Creature Academy can be used to explore evolutionary mechanisms, design, learning and other processes studied in artificial life simulations. Our system allows to perform hierarchical evolutionary experiments and ecosystem-inspired setups to investigate bodied creatures that interact, compete, adapt, and evolve. As a first proof of concept, we use Creature Academy to evolve morphologies and motion strategies of virtual creatures that walk and jump. We then present results that compare hierarchical evolution scenarios to generate creatures that excel in both walking and jumping, demonstrating how to evolve from creature specialists to generalists.},
keywords = {artificial life, evolutionary design},
pubstate = {published},
tppubtype = {inproceedings}
}
In this paper, we present Creature Academy, a virtual laboratory that allows for the evolution of form and function within simulated physical 3D environments. Creature Academy can be used to explore evolutionary mechanisms, design, learning and other processes studied in artificial life simulations. Our system allows to perform hierarchical evolutionary experiments and ecosystem-inspired setups to investigate bodied creatures that interact, compete, adapt, and evolve. As a first proof of concept, we use Creature Academy to evolve morphologies and motion strategies of virtual creatures that walk and jump. We then present results that compare hierarchical evolution scenarios to generate creatures that excel in both walking and jumping, demonstrating how to evolve from creature specialists to generalists.