Ivan-267

Sounds interesting, especially the multi-agent support. Might need some plugin and env modifications to differentiate the different agent types, and if making an example using MAPPO, some suitable/cooperative task idea.

https://github.com/edbeeching/godot_rl_agents/assets/61947090/bc4146c6-8e78-435b-b9a3-900d453f1b9f Sharing a small test case from yesterday based on the examples from SKRL. It's PPO with num_agents set to 1 and only one observation space used, not a proper/full...

Sure, we could work on a potential more complete implementation at any point. Just asking since I'm not that familiar with any MARL implementation yet and not sure what to...

Hello and welcome! There's an old 3.5 branch here https://github.com/edbeeching/godot_rl_agents/tree/godot3.5, you could try the branch, but it may not support the newer plugin or features, and it might not be...

During training, character still sometimes fell (on uneven terrain in the test case). I added a debug message for that. Not sure whether the changes cause this, it needs some...

Hello, thanks for reporting this, I have been aware of this limitation due to the implementation, we use "obs" in all of the examples except the camera one. Also onnx...

I added the change, PR here: https://github.com/edbeeching/godot_rl_agents/pull/212 This still isn't well tested (only quickly), feel free to check if it works well with a change in the obs key. It...

> Hi, I just tested your changes. Unfortunately I still get `KeyError`. Is this with camera sensor? So far I tried with vector obs only, although it was a brief...