AAU Student Projects - visit Aalborg University's student projects portal
A master's thesis from Aalborg University
Book cover


Efficient Resource Management in StarCraft: Brood War

Authors

; ; ;

Term

1. term

Publication year

2010

Submitted on

Pages

49

Abstract

I realtidsstrategispil som StarCraft afhænger succes af en stabil strøm af ressourcer til at bygge en hær og udvide basen. Vi udvikler en algoritme, der gør indsamling af mineraler mere effektiv og mere forudsigelig end spillets indbyggede minedrift. Her betyder forudsigelig, at indkomsten varierer mindre på tværs af situationer, så planlægning bliver lettere. Vi præsenterer også foreløbige resultater for en scouting-strategi—at sende enheder ud for at udforske—til at vælge, hvor man bør udvide en base. I vores tests opnår minedriftsalgoritmen en højere indtjeningshastighed end standardtilgangen og opfører sig mere ensartet. Endelig finder vi en sammenhæng mellem egenskaber i et område på kortet og de steder, som menneskelige spillere oftere vælger at udvide til.

In real-time strategy games such as StarCraft, success depends on a steady flow of resources to build an army and expand. We develop an algorithm that makes mineral collection more efficient and more predictable than the game’s built-in mining. Here, predictable means the income varies less across situations, making planning easier. We also present preliminary results on a scouting strategy—sending units to explore—to choose where to expand a base. In tests, our mining algorithm achieves a higher income rate than the default approach and behaves more consistently. Finally, we find a correlation between features of a map area and the expansion locations human players are more likely to choose.

[This abstract was generated with the help of AI]

Other projects by the authors