Fix action assignment bug in multi-dimensional action space#5889
Closed
mortalreminderpt wants to merge 3 commits intoUnity-Technologies:developfrom
Closed
Fix action assignment bug in multi-dimensional action space#5889mortalreminderpt wants to merge 3 commits intoUnity-Technologies:developfrom
mortalreminderpt wants to merge 3 commits intoUnity-Technologies:developfrom
Conversation
Modify self._current_action[current_behavior] so that actions can be passed to Unity exactly.
MultiDiscrete actions in Unity cannot be executed correctly, so you need to assign action.discrete directly to self._current_action[current_behavior].discrete
verified that it can correctly assign actions in each case
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Proposed change(s)
This PR fixes a bug that causes all actions to be assigned to the first dimension in a multi-dimensional action space. I modified the action assignment logic to handle different types of action spaces, such as discrete, continuous, and hybrid. I tested my code on three environments: SoccerTwo, FoodCollector, and 3DBall, and verified that it can correctly assign actions in each case.
Useful links (Github issues, JIRA tickets, ML-Agents forum threads etc.)
Types of change(s)
Checklist
Other comments