Reinforcement Learning Project with multiple BlueROV2s at indoor pool

Howdy Fellow Marine Explorers,

BLUF: Conducting reinforcement learning with BlueROV2 for formation control.

Background: As a current grad student that is studying reinforcement learning with multi-rotor UAVs I want to start a project using AUVs. My professor has three BlueROV2s that are not being used and I want to train one to two that follow an AUV or target in the water using their camera (don’t have funds to purchase that sweet sonar yet). We’re using gymnasium environments currently that go in hand with ray[rllib] and have looked at other RL modules that interact with python language.

Is there any public repository for dynamics and/or navigation software?
Has work like this already occurred with any BlueRobototics hardware in the past?