Gymnasium api - #1
Open
korneelf1 wants to merge 76 commits into
Open
Conversation
…sition controller
…oneSim_gym into gymnasium_env
…actions, try different collector, skip action weight (not even sure what it should be as we use default drone)
…east 3M interactions (see learning to fly)
…t to impossible states
FEATURE: time limit in done function
Author
|
As for the Kerneller and Jitter, now created in gym_sim |
Author
|
Adding basic unit tests for ensuring same output as og sim |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This pull request adds a file which adds functionalities to the simulator such as:
Offers a step function which operates as expected (performs one singular step for N_drones)
When using it with conventional RL frameworks, you should stick to N_drones = 1.
When using N_drones > 1, apply a post processor to add the transitions to the buffer for each drone (depending on how your framework operates).
Offers a step_rollout function which acts similar to the step function, but accepts a controller as input to accelerate the rollout collection. While this method allows a 50x increase in speed, care should be given to eg exploration vs exploitation (no noise is injected now) and the way that the resets are handled for each independent requirement (see reset_subenvs).
TODO:
reset sub envs using Numba, easier, faster