Desenvolvimento de um agente inteligente utilizando Q-learning para um ambiente de aprendizado

Abstract

With the recent advance of digital entertainment, there is a great demand in the market for developers with the ability to work with artificial intelligence in the games industry. To facilitate mastery of this area, several learning environments focused on artificial intelligence have been proposed in the literature with different levels of complexity and learning curves. In this paper, we propose the development of an artificial intelligence learning environment in the Python language and with a focus on beginners. The proposed environment offers a simulation of a cops and robbers game with a simple set of rules and an easy-to-learn programming interface. In addition, we also propose an implementation of an intelligent agent to validate the environment, trained with the Q-Learning method. The experimental results demonstrate that this training methodod is effective for creating and testing of intelligent behavior in the proposed simulation environment.