[Proposal] Add wind attribute for stochastic environments

[pseudo-rnd-thoughts]

In Distributional Reinforcement Learning with Quantile Regression, they propose a testing environment where wind is added to the environment to make a gridworld problem stochastic.

I propose a new environment or minigrid environment argument to add this wind to any environment I favour the new environment approach as I think this is a feature that other environment would not want or need

[siddarth-c]

I have started working on this

[siddarth-c]

Sorry for the delay! I got busy with something else. Will a wrapper work for this? Something like: It takes in a list of size [1 x grid_width] as input, where each element represents the probability of the agent being pushed up north by one square?

[pseudo-rnd-thoughts]

Yes, a wrapper would work but this would require modifying the actual environment data which a wrapper doesn't normally operate on. If you have an implementation then it shouldn't be too hard to move the implementation from a wrapper to an environment or vice-versa

[riiswa]

@siddarth-c are you still on it ?