Introducing ML-Agents

First released in September 2017, ML-Agents has rapidly evolved with the input from ML scientists, game developers, and the wider Unity fanbase due to its open source nature. This rapid progress can, at times, make it challenging to learn how to use it, with many tutorials quickly becoming outdated. However, with the release of version 1 of ML-Agents, these significant backward-incompatible updates should slow down as the project stabilizes. This means it is a great time to jump into the world of ML in Unity!

The ML-Agents toolkit consists of the following:

• The ML-Agents Unity package: This provides everything we need to implement an Agent inside the Unity environment.
• The mlagents Python package: Contains the ML algorithms that we will use to train the Agent.
• The mlagents_env Python package: Provides the functionality for Unity and the ML algorithms to talk to each other. mlagents relies on this.
• The gym_unity Python package: A wrapper...