Strategy construction using genetic algorithms for a real-time strategy game
Games are good domains for strategy and decision making problems due to its unpredictability. Real-time strategy games require decision making in every situation. However, most implementation of computer opponents use hard-coded rules, making the game repetitive and predictable. Thus, experienced hu...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | text |
| Published: |
Animo Repository
2007
|
| Subjects: | |
| Online Access: | https://animorepository.dlsu.edu.ph/faculty_research/7426 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | De La Salle University |
| Summary: | Games are good domains for strategy and decision making problems due to its unpredictability. Real-time strategy games require decision making in every situation. However, most implementation of computer opponents use hard-coded rules, making the game repetitive and predictable. Thus, experienced human players eventually learn and formulate strategies to exploit this weakness. This research has investigated the use of a machine learning algorithm, specifically Genetic Algorithm, in creating strategies for the computer opponent to achieve a degree of unpredictability in a real-time strategy game. Computer player's strategies are generated through evolution on already existing strategies. Each new strategy is given a fitness score by using them in simulations of scenarios commonly encountered in game. Evolution of the strategies is done continuously until an acceptable fitness score is achieved. Acceptable strategies are then used by the computer player in the actual game. |
|---|