In this chapter, we looked at several interesting and valuable learning environments, saw how their interfaces are set up, and even got hands-on with those environments using the quickstart guides for each environment and the setup scripts available in the book's code repository. We first looked at environments that have interfaces compatible with the OpenAI Gym interface that we are now very familiar with. Specifically in this category, we explored the Roboschool and Gym Retro environments.
We also looked at other useful learning environments that did not necessarily have a Gym-compatible environment interface, but had a very similar API and so it was easy to adapt our agent code or implement a wrapper around the learning environment to make it compatible with the Gym API. Specifically, we explored the famous real-time strategy game-based StarCraft II environment...
