DCA: Dynamic Challenging Level Adapter for Real-time Strategy Games

Abstract

Recently, Real-Time Strategy (RTS) games, such as Star Craft and Age of Empire, become more and more popular. Reasons of these RTS games attracting many game players are not only the fancy game presentation but also challenging game AI of computer opponents. In order to match game challenging level to different game players, these RTS games always provide several default difficulty levels for game players' choosing. However, settings of these difficulty levels cannot always accommodate challenging level requirements of different level game players. Therefore, this paper proposes a Dynamic Challenging Level Adapter (DCA) mechanism to automatically adapt computer opponent's behaviors for different game players. Each game player doesn't need to choose a difficulty level in advance and the game AI of computer opponent controlled by the DCA mechanism can dynamically adapt for meeting the challenging level of each game player. This paper proposes that the warrior is the most important element to dominate the result of each match in RTS games. Therefore, the main idea underlying the DCA mechanism is based on analyzing warrior capabilities between two game players and real-time adjusting action strategies to fight with its opponent. In order to test the effectiveness of the DCA mechanism, this study applies the most popular RTS game, Star Craft II (SC II), as the experimental game platform. In the experiments, this study uses default difficulty level AI of SC II game to emulate game players and makes these emulated game players to fight with the opponents controlled by the DCA mechanism. Additionally, a challenging rate (CR) formula is proposed as a strength evaluation between two opponents. From the experimental results, the difference of CR value of computer opponent improved by the DCA mechanism can be reduced by more than 80%. Additionally, the duration of a game match is usually double. Furthermore, the winner is always decided in the rear of each game match.