OpenTrack

GALBOT · Tsinghua

📃Paper | 🏠Website

This repository is the official implementation of OpenTrack, an open-source humanoid motion tracking codebase that uses MuJoCo JAX for simulation and supports multi-GPU parallel training.

News 🚩

[November 30, 2025] LAFAN1 generalist v1 released. Now you can track cartwheel, kungfu, fall and getup, and many other motions within a single policy.

[September 19, 2025] Simple Domain Randomization released.

[September 19, 2025] Tracking codebase released.

TODOs

• Release motion tracking codebase
• Release simple domain randomization
• Release pretrained LAFAN1 generalist v1 checkpoints
• Release DAgger code
• Release AnyAdapter
• Release more pretrained checkpoints
• Release real-world deployment code

Prepare

1. Clone the repository:

   git clone git@github.com:GalaxyGeneralRobotics/OpenTrack.git

2. Create a virtual environment and install dependencies:

   uv sync -i https://pypi.org/simple 

3. Create a .env file in the project directory with the following content:

   export GLI_PATH=<your_project_path>
   export WANDB_PROJECT=<your_project_name>
   export WANDB_ENTITY=<your_wandb_entity>
   export WANDB_API_KEY=<your_wandb_api_key>

4. Download the mocap data and put them under storage/data/mocap/. Thanks for the retargeting motions of LAFAN1 dataset from LocoMuJoCo!

   The file structure should be like:

   storage                   
   ├── assets
   │   ├── mujoco_menagerie
   │   └── unitree_g1
   ├── data
   │   ├── hfield            
   │   │   └── terrain.npz
   │   └── mocap
   │       └── lafan1   
   │           └── UnitreeG1
   ├── logs
   │   ├── dagger      # dagger student checkpoints
   │   └── track       # tracking checkpoints  
   └── ...
   

5. Initialize assets

   # First, initialize the environment
   source .venv/bin/activate; source .env;
   # Then, initialize the assets
   python track_mj/app/mj_playground_init.py

Usage

Initialize environment

  source .venv/bin/activate; source .env;

Play pretrained checkpoints

1. Download pretrained checkpoints and configs from checkpoints and configs, and put them under storage/logs/dagger. Visualization results: videos.

2. Run the evaluation script:

   python -m track_mj.eval.dagger.mj_onnx_video --task G1TrackingGeneral --exp_name general_tracker_lafan_v1 [--use_viewer] [--use_renderer] [--play_ref_motion]

As of November 30, 2025, we have open-sourced a generalist model on LAFAN1, daggered from four teachers. This checkpoint was trained with simple domain randomization (DR). You may try deploying it on a Unitree G1 robot using your own deployment code, since we have not yet open-sourced our real-robot deployment pipeline.

Train from scratch

[Optional] Preprocess the motion data

If you want to train on your own motion data, you should first put the .npz files under storage/data/mocap/<your_dataset_name>/UnitreeG1. Then, you should run the following preprocess script to:

1. Align the frequency or the original motion data to the desired control frequency (50Hz by default).
2. Recalculate velocities (angular, linear, joint) and other state features based on the aligned frequency.

Note: The preprocess script will overwrite the original motion files.

# Use `num_batches` to split data into multiple batches for parallel processing on multiple GPUs. You should manually launch multiple processes on different GPUs for parallel processing.
python scripts/process_motion/preprocess_motion.py --task G1TrackingGeneral --num_batches XXX --smooth_start_end False

# Or run on a single GPU without parallelism
python scripts/process_motion/preprocess_motion.py --task G1TrackingGeneral --num_batches 1 --smooth_start_end False

Argument --smooth_start_end True can generate a natural transition motion from the default pose before the original motion.

An inverse-kinematics solver generates a smooth stepping motion from the default pose to the first motion frame. The solver uses QP optimization (OSQP) with foot position/orientation constraints and CoM balance constraints. One or two foot steps are automatically chosen based on the foot-placement gap between default pose and the pose of the first frame.

Here is a comparison of the original motion and the smoothed motion:

Original Motion	Smoothed Motion

Train the specialist teachers

python -m track_mj.learning.train.train_ppo_track --task G1TrackingGeneralDR --exp_name <your_exp_name>

If you want to modify the training configuration, you should edit the config file associated with the corresponding environment. For example, for the command above, you should modify the g1_tracking_general_dr_task_config in the OpenTrack/track_mj/envs/g1_tracking/train/g1_env_tracking_general_dr.py.

Evaluate the model

Evaluate the specialist teachers

# First, convert the Brax model checkpoint to ONNX
python -m track_mj.app.brax2onnx_tracking --task G1TrackingGeneral --exp_name <your_exp_name>

# Next, run the evaluation script
python -m track_mj.eval.tracking.mj_onnx_video --task G1TrackingGeneral --exp_name <your_exp_name> [--use_viewer] [--use_renderer] [--play_ref_motion]

Acknowledgement

This repository is build upon jax, brax, loco-mujoco, and mujoco_playground.

If you find this repository helpful, please cite our work:

@misc{zhang2025trackmotionsdisturbances,
      title={Track Any Motions under Any Disturbances}, 
      author={Zhikai Zhang and Jun Guo and Chao Chen and Jilong Wang and Chenghuai Lin and Yunrui Lian and Han Xue and Zhenrong Wang and Maoqi Liu and Jiangran Lyu and Huaping Liu and He Wang and Li Yi},
      year={2025},
      eprint={2509.13833},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2509.13833}, 
}