Farama-Foundation / Minigrid

Simple and easily configurable grid world environments for reinforcement learning
https://minigrid.farama.org/
Other
2.09k stars 604 forks source link

the concrete meaning of obs #191

Closed initial-h closed 2 years ago

initial-h commented 2 years ago

Hi, I'm running experiments on env RedBlueDoors with FullyObsWrapper and ImgObsWrapper. There are three 2D arrays to present the features. I think the first array presents objects ID, the second one presents color ID and the third one presents agent's direction, it this correct? I find in RedBlueDoors, agent's direction and the status of door (open/closed) are all presented in the third array. I wonder is there a document that explains which array present what kind of features? Thanks!

maximecb commented 2 years ago

Hello.

The first array is the object type, the second is the color, and the third is a flag indicating for example if a door is open or not: https://github.com/Farama-Foundation/gym-minigrid/blob/master/gym_minigrid/minigrid.py#L238

initial-h commented 2 years ago

Thank you very much! And the agent's direction is also presented in the third array right? I'm trying to find some logic that what kind of features are put together :)