Reinforcement Learning as an Approach to Train Multiplayer First-Person Shooter Game Agents Artigo Académico