In recent years, Deep Reinforcement Learning has made great progress in video games, including Atari, ViZDoom, StarCraft, Dota2, and so on. Those successes coupled with the release of the ML-Agents Toolkit, an open-source that helps users to create simulated environments, shows that Deep Reinforcement Learning can now be easily apply to video games. Therefore, stimulating the creativity of developers and researchers. This research aspires to develop a new video game and turn it into a simulation environment for training intelligent agents. Experienced it with tuning the hyperparameters to make the agent getting the best performance for a final commercial video game product.