Merged
Conversation
* added several test cases for robots * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* [feature] ground and lidar * [update] doc update and drop redundant modifications
* [update] rslrl ppo extracted * [fix] change walk_g1 task to clean rl style * standarize algorithm configs * [update] Change RLTask to adapt to rsl-rl framework * [update] change saving directory * [update] doc * [fix] isaacgym import for sac * [update] refractor config system * [delete] redundant code removal
* Move unitree_rl from roboverse_learn to roboverse_pack BREAKING CHANGE: Module path changed from roboverse_learn.rl.unitree_rl to roboverse_pack.tasks.unitree_rl Files moved: - configs/ directory (all configuration files) - deploy/ directory (deployment scripts and configs) - helper/ directory (utility functions) Files updated with new import paths: - roboverse_pack/tasks/unitree_rl/base/*.py - roboverse_pack/tasks/unitree_rl/locomotion/*.py - roboverse_pack/tasks/unitree_rl/loco_manipulation/*.py - roboverse_pack/tasks/unitree_rl/configs/**/*.py - roboverse_pack/tasks/unitree_rl/deploy/**/*.py - roboverse_learn/rl/rsl_rl/eval.py - docs/source/roboverse_learn/reinforcement_learning/unitree_rl.md - metasim/sim/isaacgym/isaacgym.py Old directory removed: - roboverse_learn/rl/unitree_rl/ Migration consolidates all unitree_rl functionality under roboverse_pack.tasks for better organization. [refractor] unitree rl to humanoid [doc] update [update] doc name * add dr to collect demo * clean up code & create a manager * add DR to act eval * add DR to dp eval * [fix] move deploy to scripts to avoid import error --------- Co-authored-by: gxyes <gxynoz@gmail.com>
* clean up ACT * pre-commit
* Update pick_place configs and tasks * Fix bug that success rate is more than 100% in evaluate_lift.py; Fixed hard-coded path in track.py * add contributor * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* refactor DP/FM/VITA * pre-commit * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* Add MeanFlow, Improved MeanFlow, consistency flow * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * update readme --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* Update pick_place configs and tasks * Fix bug that success rate is more than 100% in evaluate_lift.py; Fixed hard-coded path in track.py * add contributor * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Fixing success rate bug; Enable more flexible evaluation that user can create less number of environment in evaluation * Make training script individual; Add gripper downward reward * Update pick_place configs and tasks * Fix bug that success rate is more than 100% in evaluate_lift.py; Fixed hard-coded path in track.py * Fixing success rate bug; Enable more flexible evaluation that user can create less number of environment in evaluation * Make training script individual; Add gripper downward reward * Resolved merge conflicts by keeping local versions * "Refactor approach_grasp tasks, yaml, and track tasks yaml" * "Refactor pick_place grasp logic and add object configs" * Refactor pick_place grasp logic and add object configs * Adjust Spoon, two spoon class in one file, two spoon yaml. * Apply pre-commit fixes and add new pick_place_spoon2 config * Local devcontainer config (ignored from push) * Fix duplicated sim key in pick_place_spoon2.yaml * Trigger CI * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * update the position of banana and screw driver * add approach_grasp.py * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Roll back readme * roll back * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * roll back collect demo * roll back collect demo * Revise devcontainer.json for development setup Updated the devcontainer configuration with new settings, including service name, workspace folder, and VSCode customizations. * Remove devcontainer.json from .gitignore --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by: Anchor1021 <bzhao78@wisc.edu>
* Fix GS background blending: exclude terrain from foreground mask using instance segmentation * Update * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Fix * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Remove * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Fix mojuco gs render * Update * Update * Lint * Update * Upate * Update * [fix] mjcf scale * Update * support rerun visualizer * update rerun readme, doc and index * update rerun code format to fit pre-commit hook * unified viz wrapper * Remove deps on pyquaternion --------- Co-authored-by: nemo.liu <nemo.liu@gpu-4090-dev015.hogpu.cc> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by: MurphyZhao04 <xh930717304@gmail.com> Co-authored-by: geng-haoran <ghr@berkeley.edu>
* [fix] unify act path with dp * [fix] unify act path with dp * [fix] act run control * [fix] act output path
I believe the change only effects Calvin tasks, thus no need to do any test. * add success check * Refactor BaseCalvinTableTask_A and clean up code * Update scene_B.py * add success check * Update scene_D.py * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * add success check * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Update scene_A.py * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Update scene_B.py * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Update scene_C.py * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Update scene_D.py * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by: 朱道远 <darwin@zhudaoyuandeMacBook-Air.local>
* [fix] VP-score matching * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* [fix] act sr stabilization * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* add two task * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * add two task * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * add two task * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * add two task * add two task --------- Co-authored-by: 朱道远 <darwin@zhudaoyuandeMacBook-Air.local> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* [fix] camera pos with env origin * [fix] replace self.sim.reset() with more accurate sensor init
…738) * [fix] mjcf scale * support rerun visualizer * update rerun readme, doc and index * update rerun code format to fit pre-commit hook * unified viz wrapper * [update] autotest trigger for forked repo (#713) * [update] autotest trigger for forked repo * [fix] drop declared core * Dev/fork autotest (#715) * [update] autotest trigger for forked repo * [fix] drop declared core * [fix] drop cli action * [fix] use pat token for triggering autotest (#716) * [fix] use pat token for triggering autotest * [update] merge priv-ci into merge queue * [fix] secrets grammar error * Add support for 6 dexterous hands with demo script and documentation This commit introduces support for three types of dexterous hands (6 variants total): **Inspire Hand (Left/Right)** - 12 DoFs total: 6 actuated joints + 6 mimic/passive (coupled) joints - Thumb: 2 actuated (yaw + pitch) + 2 passive (intermediate, distal) - Four fingers: 1 actuated (proximal) + 1 passive (intermediate) each - Passive joints mechanically follow actuated joints via coupling **BrainCo Hand (Left/Right)** - 11 DoFs total: 6 actuated joints + 5 mimic/passive joints - Prosthetic hand design with coupled joint mechanisms - 6 independent control inputs drive all finger movements **PSIHand (Left/Right)** - 21 DoFs total: All 21 joints fully actuated - 5 fingers with 4 joints each (thumb has 5 joints) - Note: Known compatibility issues with IsaacGym (NaN joint positions in URDF mode, DOF force tensor failures in USD mode). Use MuJoCo or Genesis instead. **Components Added:** - 6 robot configuration files in roboverse_pack/robots/ - Control demo script: get_started/dexhands/1_diff_dex_hand.py - Updated documentation: docs/source/dataset_benchmark/dataset/robots.md - Registered new hands in roboverse_pack/robots/__init__.py The demo script supports all RoboVerse simulators and includes camera configuration for video recording. For IsaacGym, initial pose is set to properly orient hands. --------- Co-authored-by: MurphyZhao04 <xh930717304@gmail.com> Co-authored-by: geng-haoran <ghr@berkeley.edu> Co-authored-by: Mingyuan Sun <64095400+myuansun@users.noreply.github.com> Co-authored-by: 23369 <2336988354@qq.com> Co-authored-by: Murphy <167211955+MurphyZhao04@users.noreply.github.com>
* [fix] usd_path of g1_cfg.G1Dof29Cfg * [fix] usd_path of g1_cfg.G1Dof29Cfg * Start migrating BeyondMimic * Slightly adapt unitree_rl main.py to BeyondMimic play.py (not finished) * Correct import paths * Save before adapting BeyondMimic evaluation * Start adapting BeyondMimic to unitree_rl infra * Add rewards, observations, terminations, randomization, motion commands, etc. * Ready to adapt main entry * Almost finished implementation * Debug and update * Able to run but metrics are not right * Fix a bug of mean reward * Able to run evaluation script but motions are probably not loaded correctly * Fix bugs about joint and body orders * Fix a bug of joint order of applied actions * Add G1 robot config from BeyondMimic * Debug soft_joint_pos_limit_factor * Adding Isaac Lab version BeyondMimic * Able to run Isaac Lab-specific BeyondMimic training * Implement training wrappers for both RSL v2.3.0 and v3.1.0; fix issues about entropy_coef * Save before debugging motion tracking evaluation pipeline * Refactor BeyondMimic MetaSim version * Refactor and debug task class; about to adapt evaluation script * Finish unified evaluation script for tracking; save before toggling __init__.py of submodules * Able to evaluate but previous issues are not fixed * Fix a bug in * Fix issues about updating commands after first reset and resampling commands; evaluation should be alright now * Fix a small issue of dimension mismatch in command resampling * Fix a bug of RoboVerse env wrapper; add a task class for training tracker for deployment * Clean up roboverse_pack * Clean up roboverse_learn, add README under docs/source/roboverse_learn/reinforcement_learning, and add scripts for converting motions * Modify BeyondMimic README * Clean up tracking_g1 * Try to re-merge * Clean up comments * Add URDF to USD conversion script; remove robot instantiation using URDF config in Isaac Sim handler * [fix] isaacgym gs * [fix] env origin * [update] move math functions into metasim.utils.math * [update] premerge-ci --------- Co-authored-by: myuansun <1273414643@qq.com>
* [fix] effort limit * [fix] usd path issue * [update] premerge-ci
Co-authored-by: boshi-an <boshi_an2stu.pku.edu.cn>
* [fix] adapative terrain size to num_envs * [update] simplify code * [fix] remove default lidar sensor * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: OMNILACRIM <example@mail.com> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* [fix] wandb logging * [fix] optimize contact sensor & fix g1_dof12 observation buffer * [fix] fix isaacgym CUDA error * [update] only set collapse_fixed_joints to True for isaacgym * [fix] align checkpoint directory of eval.py * [fix] obs_groups & collapse fixed joints for g1 * [fix] isaacgym render with redundant physics step * [fix] add ground padding for isaacsim * [update] increase ground padding of isaacsim * [update] change default obs_group in eval.py
* feat:adpat agibot a2 dof12 into unitree rl locomotion * fix:1.Modify the model training path2.Fix device settings and model weights loading in the eval file * fix:1.Modify the model training path2.Fix device settings and model weights loading in the eval file * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Fix: add missing docstrings * fix: reworked the logic for saving path * fix: update task registration identifier for WalkAgibotA2Dof12 * fix: modify evaluation script for Mujoco compatibility and performance logging - Added time tracking for evaluation steps to monitor simulation and elapsed time. - Adjusted device settings to default to CPU when using Mujoco. - Removed unnecessary observation device transfer function to streamline the evaluation process. * fix: update actuator configuration and enable USD path for AgibotA2Dof12 - Changed actuator configuration to use 'effort_limit_sim' instead of 'torque_limit'. - Enabled the USD path for the AgibotA2Dof12 robot configuration. * feat: add joint effort limits function and adjust reward scales for WalkAgibotA2Dof12 - Implemented a new function to penalize joint efforts exceeding torque limits. - Updated reward scales for linear and angular velocity, joint acceleration, action rate, and added joint effort limits to enhance locomotion performance. - Adjusted target height for stability in the environment configuration. * correct contributors in alphabetical order * Reorder Yikai Tang in contributors list * accept develop branch change --------- Co-authored-by: wangbei <wangbei@agibot.com> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by: zhangyi <zhangyi_1357@sjtu.edu.cn> Co-authored-by: Mingyuan Sun <64095400+myuansun@users.noreply.github.com>
* [add] initial newton handler supporting get started 0 * [fix] add GUI viewer code * [add] robot control * [update] add env spacing and njmax, nconmax to pass get started 0-3 * [fix] camera support env origin * [update] g1 sim2sim ready, but looks quite slow * [add] pytest for newton * [add] per-body gravity compensation * [fix] per-object joint name resolution, skip joint limit test for newton * [fix] decouple camera from OpenGL GUI, align urdf loading details and effort control details * [fix] validate inertia * [fix] contact sensor layout * [update] drop dedundant test case * [update] set newton_use_mujoco_contacts to false for training in newton * [update] add mass and material randomizer * [update] optimize training speed and fix a weird contact bug * docs: add newton support and compressed assets
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 902c65614f
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
* [fix] deploy path * [fix] import
Deploying roboverse-release with
|
| Latest commit: |
4202358
|
| Status: | ✅ Deploy successful! |
| Preview URL: | https://bcf315f6.roboverse-release.pages.dev |
| Branch Preview URL: | https://develop.roboverse-release.pages.dev |
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
14 participants
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.
Description
Merge develop into main
Type of change
Checklist
pre-commitchecks withpre-commit run --color=always --all-filesCONTRIBUTORS.mdor my name already exists there