It likely started out as a fun experiment, but as it turns out, letting agents compete against themselves can really amp up training, and, well, it's just cool. There are a few configurations we can set up when working with multiple agents. The BananaCollector example we will look at uses a single brain shared among multiple competing agents. Open up Unity and follow this exercise to set up the scene:
- Load the BananaCollectorBananaRL scene file located in the Assets/ML-Agents/Examples/BananaCollectors/ folder.
- Leave the Brain on Player; if you changed it, change it back.
- Run the scene in Unity. Use the WASD keys to move the agent cubes around the scene and collect bananas. Notice how there are multiple cubes responding ...