Reinforcement Learning in Game Engines

Reinforcement Learning in Game Engines

Sri Ganesh, Bharat Kotwani, Ajay Ganesh Suryawanshi , Kamal Mandava, Abhay Bansod, Ushasree Tella, Lanka Sai Krishna Chaitanya, Sujay Bhagawan Ghadge, Gaurav Kakoti, Deepak Kari

Abstract
Machine Learning has developed in many ways during the recent years. One of the new fields which has potential applications in machine learning is computer games. With this research I aim to develop machine learning functionalities in computer games using a lightweight 2D - Game Engine developed in Java. The engine will have the basic functionalities of any Game Engine and we will be able to use it to create simple neural networks and train these neural networks for use in the game.

The neural network will be trained using data collected from the game and we have expectations that the model will even be able to learn how to play the game by itself.

For the development of the Game Engine we will be using tools and libraries for Java for example, LWJGL 3, OpenGL, Assimp, Maven for Research Management and Research Structure Management, GLFW for event listeners, GLSL for Shaders, using ImGUI for Graphical user interface development. Serialization and Deserialization with GSON to save our designed levels. Gradle is used for Dependency and Library Management, Launch4j for converting the Packaged .jar file into an executable for Distribution.