This repository contains a Reinforcement Learning environment for Pokémon battles.
In particular, the environment consists of three parts:
- A Gym Env which serves as interface between RL agents and battle simulators
- A BattleSimulator base class, which handles typical Pokémon game state
- Simulator classes derived from BattleSimulator, which access and interact with different simulators to extract data
Currently, only a Pokemon Showdown integration is planned, but in theory this structure allows for integrations with different simulators (eg Console emulators).