[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-72608":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":8,"htmlUrl":8,"language":9,"languages":8,"totalLinesOfCode":8,"stars":10,"forks":11,"watchers":12,"openIssues":13,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":15,"stars7d":16,"stars30d":17,"stars90d":14,"forks30d":14,"starsTrendScore":18,"compositeScore":19,"rankGlobal":8,"rankLanguage":8,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":21,"hasPages":21,"topics":23,"createdAt":8,"pushedAt":8,"updatedAt":24,"readmeContent":25,"aiSummary":26,"trendingCount":14,"starSnapshotCount":14,"syncStatus":27,"lastSyncTime":28,"discoverSource":29},72608,"whole_body_tracking","HybridRobotics\u002Fwhole_body_tracking","HybridRobotics",null,"Python",2136,294,13,25,0,10,37,91,30,29.41,"MIT License",false,"main",[],"2026-06-12 02:03:05","# BeyondMimic Motion Tracking Code\n\n[![IsaacSim](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FIsaacSim-4.5.0-silver.svg)](https:\u002F\u002Fdocs.omniverse.nvidia.com\u002Fisaacsim\u002Flatest\u002Foverview.html)\n[![Isaac Lab](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FIsaacLab-2.1.0-silver)](https:\u002F\u002Fisaac-sim.github.io\u002FIsaacLab)\n[![Python](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-3.10-blue.svg)](https:\u002F\u002Fdocs.python.org\u002F3\u002Fwhatsnew\u002F3.10.html)\n[![Linux platform](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fplatform-linux--64-orange.svg)](https:\u002F\u002Freleases.ubuntu.com\u002F20.04\u002F)\n[![pre-commit](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpre--commit-enabled-brightgreen?logo=pre-commit&logoColor=white)](https:\u002F\u002Fpre-commit.com\u002F)\n[![License](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-MIT-yellow.svg)](https:\u002F\u002Fopensource.org\u002Flicense\u002Fmit)\n\n[[Website]](https:\u002F\u002Fbeyondmimic.github.io\u002F)\n[[Arxiv]](https:\u002F\u002Farxiv.org\u002Fabs\u002F2508.08241)\n[[Video]](https:\u002F\u002Fyoutu.be\u002FRS_MtKVIAzY)\n\n## Overview\n\nBeyondMimic is a versatile humanoid control framework that provides highly dynamic motion tracking with the\nstate-of-the-art motion quality on real-world deployment and steerable test-time control with guided diffusion-based\ncontrollers.\n\nThis repo covers the motion tracking training in BeyondMimic. **You should be able to\ntrain any sim-to-real-ready motion in the LAFAN1 dataset, without tuning any parameters**.\n\nFor sim-to-sim and sim-to-real deployment, please refer to\nthe [motion_tracking_controller](https:\u002F\u002Fgithub.com\u002FHybridRobotics\u002Fmotion_tracking_controller).\n\n### Alternative Implementations\n\n- There is an alternative reproduction of BeyondMimic in [mjlab](https:\u002F\u002Fgithub.com\u002Fmujocolab\u002Fmjlab), a new Isaac Lab-style manager API powered by MuJoCo-Warp for RL and robotics research. See the implementation [here](https:\u002F\u002Fgithub.com\u002Fmujocolab\u002Fmjlab\u002Fblob\u002Fmain\u002Fsrc\u002Fmjlab\u002Ftasks\u002Ftracking\u002Ftracking_env_cfg.py).\n\n## Installation\n\n- Install Isaac Lab v2.1.0 by following\n  the [installation guide](https:\u002F\u002Fisaac-sim.github.io\u002FIsaacLab\u002Fmain\u002Fsource\u002Fsetup\u002Finstallation\u002Findex.html). We recommend\n  using the conda installation as it simplifies calling Python scripts from the terminal.\n\n- Clone this repository separately from the Isaac Lab installation (i.e., outside the `IsaacLab` directory):\n\n```bash\n# Option 1: SSH\ngit clone git@github.com:HybridRobotics\u002Fwhole_body_tracking.git\n\n# Option 2: HTTPS\ngit clone https:\u002F\u002Fgithub.com\u002FHybridRobotics\u002Fwhole_body_tracking.git\n```\n\n- Pull the robot description files from GCS\n\n```bash\n# Enter the repository\ncd whole_body_tracking\n# Rename all occurrences of whole_body_tracking (in files\u002Fdirectories) to your_fancy_extension_name\ncurl -L -o unitree_description.tar.gz https:\u002F\u002Fstorage.googleapis.com\u002Fqiayuanl_robot_descriptions\u002Funitree_description.tar.gz && \\\ntar -xzf unitree_description.tar.gz -C source\u002Fwhole_body_tracking\u002Fwhole_body_tracking\u002Fassets\u002F && \\\nrm unitree_description.tar.gz\n```\n\n- Using a Python interpreter that has Isaac Lab installed, install the library\n\n```bash\npython -m pip install -e source\u002Fwhole_body_tracking\n```\n\n## Motion Tracking\n\n### Motion Preprocessing & Registry Setup\n\nIn order to manage the large set of motions we used in this work, we leverage the WandB registry to store and load\nreference motions automatically.\nNote: The reference motion should be retargeted and use generalized coordinates only.\n\n- Gather the reference motion datasets (please follow the original licenses), we use the same convention as .csv of\n  Unitree's dataset\n\n    - Unitree-retargeted LAFAN1 Dataset is available\n      on [HuggingFace](https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002Flvhaidong\u002FLAFAN1_Retargeting_Dataset)\n    - Sidekicks are from [KungfuBot](https:\u002F\u002Fkungfu-bot.github.io\u002F)\n    - Christiano Ronaldo celebration is from [ASAP](https:\u002F\u002Fgithub.com\u002FLeCAR-Lab\u002FASAP).\n    - Balance motions are from [HuB](https:\u002F\u002Fhub-robot.github.io\u002F)\n\n\n- Log in to your WandB account; access Registry under Core on the left. Create a new registry collection with the name \"\n  Motions\" and artifact type \"All Types\".\n\n\n- Convert retargeted motions to include the maximum coordinates information (body pose, body velocity, and body\n  acceleration) via forward kinematics,\n\n```bash\npython scripts\u002Fcsv_to_npz.py --input_file {motion_name}.csv --input_fps 30 --output_name {motion_name} --headless\n```\n\nThis will automatically upload the processed motion file to the WandB registry with output name {motion_name}.\n\n- Test if the WandB registry works properly by replaying the motion in Isaac Sim:\n\n```bash\npython scripts\u002Freplay_npz.py --registry_name={your-organization}-org\u002Fwandb-registry-motions\u002F{motion_name}\n```\n\n- Debugging\n    - Make sure to export WANDB_ENTITY to your organization name, not your personal username.\n    - If \u002Ftmp folder is not accessible, modify csv_to_npz.py L319 & L326 to a temporary folder of your choice.\n\n### Policy Training\n\n- Train policy by the following command:\n\n```bash\npython scripts\u002Frsl_rl\u002Ftrain.py --task=Tracking-Flat-G1-v0 \\\n--registry_name {your-organization}-org\u002Fwandb-registry-motions\u002F{motion_name} \\\n--headless --logger wandb --log_project_name {project_name} --run_name {run_name}\n```\n\n### Policy Evaluation\n\n- Play the trained policy by the following command:\n\n```bash\npython scripts\u002Frsl_rl\u002Fplay.py --task=Tracking-Flat-G1-v0 --num_envs=2 --wandb_path={wandb-run-path}\n```\n\nThe WandB run path can be located in the run overview. It follows the format {your_organization}\u002F{project_name}\u002F along\nwith a unique 8-character identifier. Note that run_name is different from run_path.\n\n## Code Structure\n\nBelow is an overview of the code structure for this repository:\n\n- **`source\u002Fwhole_body_tracking\u002Fwhole_body_tracking\u002Ftasks\u002Ftracking\u002Fmdp`**\n  This directory contains the atomic functions to define the MDP for BeyondMimic. Below is a breakdown of the functions:\n\n    - **`commands.py`**\n      Command library to compute relevant variables from the reference motion, current robot state, and error\n      computations. This includes pose and velocity error calculation, initial state randomization, and adaptive\n      sampling.\n\n    - **`rewards.py`**\n      Implements the DeepMimic reward functions and smoothing terms.\n\n    - **`events.py`**\n      Implements domain randomization terms.\n\n    - **`observations.py`**\n      Implements observation terms for motion tracking and data collection.\n\n    - **`terminations.py`**\n      Implements early terminations and timeouts.\n\n- **`source\u002Fwhole_body_tracking\u002Fwhole_body_tracking\u002Ftasks\u002Ftracking\u002Ftracking_env_cfg.py`**\n  Contains the environment (MDP) hyperparameters configuration for the tracking task.\n\n- **`source\u002Fwhole_body_tracking\u002Fwhole_body_tracking\u002Ftasks\u002Ftracking\u002Fconfig\u002Fg1\u002Fagents\u002Frsl_rl_ppo_cfg.py`**\n  Contains the PPO hyperparameters for the tracking task.\n\n- **`source\u002Fwhole_body_tracking\u002Fwhole_body_tracking\u002Frobots`**\n  Contains robot-specific settings, including armature parameters, joint stiffness\u002Fdamping calculation, and action scale\n  calculation.\n\n- **`scripts`**\n  Includes utility scripts for preprocessing motion data, training policies, and evaluating trained policies.\n\nThis structure is designed to ensure modularity and ease of navigation for developers expanding the project.\n","BeyondMimic 是一个灵活的人形控制框架，专注于提供高质量的全身运动跟踪。项目利用先进的引导扩散控制器实现动态运动跟踪，并支持无需参数调整即可训练LAFAN1数据集中的任意仿真到现实的运动。技术上，它基于IsaacSim 4.5.0和Isaac Lab 2.1.0开发，使用Python 3.10编写，运行于Linux 64位平台。此外，项目还提供了对MuJoCo-Warp的支持以增强其在强化学习与机器人研究中的应用。适用于需要高精度运动捕捉及控制的研究场景，如机器人学、虚拟现实以及动画制作等领域。",2,"2026-06-11 03:42:46","high_star"]