Repository navigation
refactor: extract shared locomotion reward functions (#149) - #183
Merged
Merged
Conversation
Introduce RewardContext dataclass and 16 shared reward functions that joystick environments reference directly in their _reward_fns dispatch tables, eliminating ~38 duplicate _reward_* methods across G1/Go1/Go2.
2 tasks done
TATP-233
added a commit
that referenced
this pull request
May 4, 2026
refactor: extract shared locomotion reward functions (#149)
TATP-233
added a commit
that referenced
this pull request
May 30, 2026
refactor: extract shared locomotion reward functions (#149)
TATP-233
added a commit
that referenced
this pull request
Sep 4, 2026
refactor: extract shared locomotion reward functions (#149)
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
RewardContextdataclass incommon/rewards.pybundling all state reward functions needtracking_lin_vel,action_rate,base_height, etc.) as standalonefn(ctx: RewardContext) -> np.ndarray_reward_fnsdispatch tables — no wrapper methods_reward_*methods + 3 module-level helpers across G1/Go1/Go2ctxsignature_reward_fnsdict has the exact same keys as beforeTest plan
make check(ruff format + ruff check + mypy + pyright) — all passmake test— 329 passed, 12 skippeduv run python scripts/train_rsl_rl.py task=go2_joystick algo.max_iterations=3— reward curves unchangedCloses #149