DocumentCode :
1286033
Title :
The evolution of swarm grammars- growing trees, crafting art, and bottom-up design
Author :
von Mammen, Sebastian ; Jacob, Christian
Author_Institution :
Univ. of Calgary, Calgary, AB, Canada
Volume :
4
Issue :
3
fYear :
2009
Firstpage :
10
Lastpage :
19
Abstract :
We presented swarm grammars as an extension of Lindenmayer systems. Instead of applying a single (\´turtle\´) agent to convert linear strings into 3D structures, we use a swarm of agents "which navigate in 3D space and-as a side effect-place structural building blocks into their environment. The swarm grammars are used to specify how the setup of agent types changes over time. Additional agent parameters determine the agents\´ behaviors and their interaction dynamics. Both the grammar rules and the agent parameters are evolvable and can change over time-either automatically at replication and collision events among the agents, or triggered by external \´tinkering\´ from a supervising breeder. When swarm grammars are applied to concrete problems, constraints on the developmental processes as "well as on the emerging structures may provide the basis for an automatic evolutionary algorithm.
Keywords :
evolutionary computation; grammars; multi-agent systems; 3D structures; Lindenmayer systems; automatic evolutionary algorithm; multiagent systems; swarm grammars; Art; Biological system modeling; Bridges; Buildings; Embryo; Environmental factors; Jacobian matrices; Multiagent systems; Neural networks; Organisms;
fLanguage :
English
Journal_Title :
Computational Intelligence Magazine, IEEE
Publisher :
ieee
ISSN :
1556-603X
Type :
jour
DOI :
10.1109/MCI.2009.933096
Filename :
5190930
Link To Document :
بازگشت