In this chapter, we took a short tour of many basic concepts involving your next steps in DL and DRL; perhaps you will decide to pursue the Unity Obstacle Tower Challenge and complete that or just use DRL in your own project. We looked at simple quizzes in order to evaluate your potential for diving in and using DRL in a game. From there, we looked at the next steps in development, and then finally we looked at other areas of learning may want to focus on.

This book was an exercise in understanding how effective DL can be when applied to your game project in the future. We explored many areas of basic DL principles early on and looked at more specific network types such as CNN and LSTM. Then, we looked at how these basics network forms could be applied to applications for driving and building a chatbot. From there, we looked at the current king of machine learning algorithms, reinforcement and deep reinforcement learning. We then looked at one of the current leaders, Unity ML-Agents, and how to implement this technology, over several chapters by looking at how simple environments are built to more complex multi-agent environments. This also allowed us to explore different forms of intrinsic/extrinsic rewards and learning systems, including curriculum, curiosity, imitation, and transfer learning.

Finally, before finishing this chapter, we completed a long exercise regarding using DRL for automatic testing and debugging with the added option of using IL as a way of enhancing testing.