Adaptive Strategy Decision Mechanism forStarCraft AI

摘要

Recent advancement of computer systems and their global networks have enabled us to enjoy many kinds of on-line games. StarCraft has been the one of the well-known on-line strategy simulation games, since it was made in 12 years ago. The game users have been extremely improved their playing level, but the AI (artificial intelligence) system of the game was rarely updated. At the consequence, current AI level is much poorer than that of the human players. In this paper, we present a noble AI mechanism for StarCraft based on the adaptive strategy decision mechanism. We implemented the proposed mechanism on the real StarCraft: BroodWar binary code. We compared the performance of the modified AI code, and the results show that the new AI code outperforms the existing artificial intelligence of the game.