Co-evolving influence map tree based strategy game players

Chris Miles; Juan Quiroz; Ryan Leigh; Sushil J. Louis

doi:10.1109/CIG.2007.368083

Co-evolving influence map tree based strategy game players

Chris Miles^*, Juan Quiroz, Ryan Leigh, Sushil J. Louis

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

38 Citations (Scopus)

Abstract

We investigate the use of genetic algorithms to evolve AI players for real-time strategy games. To overcome the knowledge acquisition bottleneck found in using traditional expert systems, scripts, or decision trees we evolve players through co-evolution. Our game players are implemented as resource allocation systems. Influence map trees are used to analyze the game-state and determine promising places to attack, defend, etc. These spatial objectives are chained to non-spatial objectives (train units, build buildings, gather resources) in a dependency graph. Players are encoded within the individuals of a genetic algorithm and co-evolved against each other, with results showing the production of strategies that are innovative, robust, and capable of defeating a suite of hand-coded opponents.

Original language	English
Title of host publication	2007 IEEE Symposium on Computational Intelligence and Games
Publisher	Institute of Electrical and Electronics Engineers (IEEE)
Pages	88-95
Number of pages	8
ISBN (Print)	9781424407095
DOIs	https://doi.org/10.1109/CIG.2007.368083
Publication status	Published - 2007
Externally published	Yes
Event	IEEE Symposium on Computational Intelligence and Games - Honolulu Duration: 1 Apr 2007 → 5 Apr 2007

Conference

Conference	IEEE Symposium on Computational Intelligence and Games
City	Honolulu
Period	1/04/07 → 5/04/07

Keywords

co-evolution
game AI
computer game
real-time strategy games

Access to Document

10.1109/CIG.2007.368083

Cite this

@inproceedings{833f1d5db8cc4cba8d19253f9314db84,

title = "Co-evolving influence map tree based strategy game players",

abstract = "We investigate the use of genetic algorithms to evolve AI players for real-time strategy games. To overcome the knowledge acquisition bottleneck found in using traditional expert systems, scripts, or decision trees we evolve players through co-evolution. Our game players are implemented as resource allocation systems. Influence map trees are used to analyze the game-state and determine promising places to attack, defend, etc. These spatial objectives are chained to non-spatial objectives (train units, build buildings, gather resources) in a dependency graph. Players are encoded within the individuals of a genetic algorithm and co-evolved against each other, with results showing the production of strategies that are innovative, robust, and capable of defeating a suite of hand-coded opponents.",

keywords = "co-evolution, game AI, computer game, real-time strategy games",

author = "Chris Miles and Juan Quiroz and Ryan Leigh and Louis, {Sushil J.}",

year = "2007",

doi = "10.1109/CIG.2007.368083",

language = "English",

isbn = "9781424407095",

pages = "88--95",

booktitle = "2007 IEEE Symposium on Computational Intelligence and Games",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

address = "United States",

note = "IEEE Symposium on Computational Intelligence and Games ; Conference date: 01-04-2007 Through 05-04-2007",

}

TY - GEN

T1 - Co-evolving influence map tree based strategy game players

AU - Miles, Chris

AU - Quiroz, Juan

AU - Leigh, Ryan

AU - Louis, Sushil J.

PY - 2007

Y1 - 2007

N2 - We investigate the use of genetic algorithms to evolve AI players for real-time strategy games. To overcome the knowledge acquisition bottleneck found in using traditional expert systems, scripts, or decision trees we evolve players through co-evolution. Our game players are implemented as resource allocation systems. Influence map trees are used to analyze the game-state and determine promising places to attack, defend, etc. These spatial objectives are chained to non-spatial objectives (train units, build buildings, gather resources) in a dependency graph. Players are encoded within the individuals of a genetic algorithm and co-evolved against each other, with results showing the production of strategies that are innovative, robust, and capable of defeating a suite of hand-coded opponents.

AB - We investigate the use of genetic algorithms to evolve AI players for real-time strategy games. To overcome the knowledge acquisition bottleneck found in using traditional expert systems, scripts, or decision trees we evolve players through co-evolution. Our game players are implemented as resource allocation systems. Influence map trees are used to analyze the game-state and determine promising places to attack, defend, etc. These spatial objectives are chained to non-spatial objectives (train units, build buildings, gather resources) in a dependency graph. Players are encoded within the individuals of a genetic algorithm and co-evolved against each other, with results showing the production of strategies that are innovative, robust, and capable of defeating a suite of hand-coded opponents.

KW - co-evolution

KW - game AI

KW - computer game

KW - real-time strategy games

U2 - 10.1109/CIG.2007.368083

DO - 10.1109/CIG.2007.368083

M3 - Conference proceeding contribution

SN - 9781424407095

SP - 88

EP - 95

BT - 2007 IEEE Symposium on Computational Intelligence and Games

PB - Institute of Electrical and Electronics Engineers (IEEE)

T2 - IEEE Symposium on Computational Intelligence and Games

Y2 - 1 April 2007 through 5 April 2007

ER -

Co-evolving influence map tree based strategy game players

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this