Book Image

Unity 2020 By Example - Third Edition

By : Robert Wells
Book Image

Unity 2020 By Example - Third Edition

By: Robert Wells

Overview of this book

The Unity game engine, used by millions of developers around the world, is popular thanks to its features that enable you to create games and 3D apps for desktop and mobile platforms in no time. With Unity 2020, this state-of-the-art game engine introduces enhancements in Unity tooling, editor, and workflow, among many other additions. The third edition of this Unity book is updated to the new features in Unity 2020 and modern game development practices. Once you’ve quickly got to grips with the fundamentals of Unity game development, you’ll create a collection, a twin-stick shooter, and a 2D adventure game. You’ll then explore advanced topics such as machine learning, virtual reality, and augmented reality by building complete projects using the latest game tool kit. As you implement concepts in practice, this book will ensure that you come away with a clear understanding of Unity game development. By the end of the book, you'll have a firm foundation in Unity development using C#, which can be applied to other engines and programming languages. You'll also be able to create several real-world projects to add to your professional game development portfolio.
Table of Contents (16 chapters)

Training the Agent

The chapter has been building up to this point. You will shortly be able to see your work up to now translate into something tangible—a pre-trained model that can be reused in any of your projects. We've created a suitable environment, populated it with a never-ending supply of opportunities for the monster to do right or wrong (by colliding with the chick or rock, respectively), and configured the learning algorithm's input and how we respond to its output. Now it is time to run the learning process, and this is where the Python packages we installed in the Installing ML-Agents section come in handy (well, essential really; without them, the monster won't be learning anything!). However, before we can run the learning command, we need to create a settings file. This file will store all the settings related to training the Agent and can have a small effect on how our monster learns:

  1. In your file browser of choice, create a folder called...