search

Panopticon-AI-team / panopticon

A wargaming platform compatible with reinforcement learning agents
https://www.panopticon-ai.com
Apache License 2.0
4 stars 0 forks source link

Research MicroRTS-Py #3

Closed duyminh1998 closed 11 months ago

duyminh1998 commented 11 months ago

We need to know more about

MicroRTS-Py, a simple and highly efficient RTS-game-inspired environment for reinforcement learning

So that

we can develop an agent for the environment and test our project

Specifically

  1. What kinds of environments/platforms does MicroRTS-Py support? (e.g. OpenAI Gym, PettingZoo, etc.)
  2. How can we run an instance of the environment? Can it be run asynchronously to our agent?
  3. What are the state variables and actions?
  4. What other agents have been developed for the environment? Which ones are the most successful?
  5. What is the best way to integrate environments like MicroRTS-Py into our own project? Do we need a wrapper? Do we need an API?
duyminh1998 commented 11 months ago

Paper

duyminh1998 commented 11 months ago

MicroRTS-Py is currently failing to run on my system. I have followed the installation instructions closely but I have failed multiple times. In the end, I do not believe the juice is worth the squeeze because whatever product (model) we create for MicroRTS-Py will not be marketable anyways. In addition, there already has been a lot of models developed for the game. Recommend we look into alternatives like AFSIM or Command.