search

geek-ai / MAgent

A Platform for Many-Agent Reinforcement Learning
MIT License
1.69k stars 332 forks source link

About modifying the number of the agents and the control of the teams? #88

Open Amanda2024 opened 3 years ago

Amanda2024 commented 3 years ago

Hi,

Taking the "battlefield" for example, how can I control the red team only? Then how can I change the number of the agents?

Thanks!

Kipsora commented 3 years ago

Please kindly refer to our docs for using our codes. You have to follow the reinforcement learning routine (i.e. (observation, action, reward) tuple) to control the agents.

Since this project is not actively maintained, I'd suggest to do some toy examples to understand each channel in the observation and how action tensor maps to the real actions in the game.