How to reset ML Agents environment when MaxStep is reached?

Unity Engine
1

Hello,

Haven’t found an answer to this, but I have a few functions that subscribe to the OnEnvironmentReset, but OnEnvironmentReset is only called at the start of my training, and never when MaxStep is reached. It feels like I’m missing something extremely simple, but I can’t find the issue.

EDIT: On further research, I understand the use cast of OnEnvironmentReset. My question now is: Is there a built in event for when MaxSteps is reached, or should I design that myself? All of my agents currently would be acting with a MaxStep of 5000, and I’d like an “OnEpisodeEnd” event to subscribe and reset my environment with.

Any help would be great!

2

I’m not sure if the ML-Agents dev team will read posts here, so you likely want to post over at Unity Engine - Unity Discussions to get more visibility.