PyTorchRL agents can be trained with any environment that complies with OpenAI gym’s interface, which allows to easily define custom environments specific to any domain of interest.
We refer here to some resources providing detailed explanations on how to implement custom environments.