search

openai / gym

A toolkit for developing and comparing reinforcement learning algorithms.
https://www.gymlibrary.dev
Other
34.67k stars 8.6k forks source link

Beginner/intermediate multi-agent grid games in ToyText #674

Closed allanbreyes closed 7 years ago

allanbreyes commented 7 years ago

Hi! Would you be interesting in us submitting a few multi-agent Markov games from literature? The motivation would be for providing more approachable alternatives to multi-agent RL other than those in the Board Games. We think that this would be a good "bridge" to go from single-agent to multi-agent RL.

Specifically, I'm proposing the grid games posed in:

As an aside, there doesn't seem to be any general-sum environments where multiple agents are controlled to achieve correlated equilibria. Would there be any interest in implementing some of these where the researcher can control multiple agents (vs. playing against a "good" adversarial agent)?

Thanks!

andertavares commented 7 years ago

Hi, I was interested in this issue, specially on Littman's soccer.

You have closed it because you have added those Markov games?

Thanks