Design and implementation of an intelligent agent for auto battler games using reinforcement learning

Abstract

This thesis explores the application of the MuZero algorithm in auto battler game. The characteristics of auto battler games are complex game mechanics, multiple players, and high randomness. The main purpose of this study is to design and implement an efficient game simulator for automatic combat games, develop an intelligent agent using the MuZero algorithm, and evaluate its performance.

Designed and implemented an efficient game simulator to provide a training environment for reinforcement learning algorithms. The developed simulator greatly reduces the time required for a single game. Using this simulator, a self-game artificial intelligence based on the MuZero algorithm begins to learn game strategies.

The main contribution of this thesis is the design and implementation of an efficient AutoBattler game simulator and the development of self-learning artificial intelligence using the MuZero algorithm. The experimental results demonstrate the effectiveness of the proposed method, providing insights for artificial intelligence development in complex gaming environments.