Sitelet https://github.com/jessicayin/osmo_tactile_glove
Skip to content

About

Open source tactile glove for robotics research

Resources

Stars

84 stars

Watchers

3 watching

Forks

Repository files navigation

OSMO: Open-Source Tactile Glove for Human-to-Robot Skill Transfer

Jessica Yin, Haozhi Qi*, Youngsun Wi*, Sayantan Kundu, Mike Lambeta, William Yang, Changhao Wang, Tingfan Wu, Jitendra Malik, and Tess Hellebrekers

*equal contribution

Project website: jessicayin.github.io/osmo_tactile_glove

ArXiv: https://arxiv.org/abs/2512.08920

OSMO Hardware

Hardware guides and PCB files are on the website branch of this repo.

OSMO Glove Data Pipeline

Code has references to "bowie", which was the internal name of OSMO. Bowie and the OSMO glove may be treated as interchangeable when understanding the code.

  1. Extract the hamer keypoints: (in osmo env)
conda activate osmo
python data_collect/glove/labs/glove2robot/postprocess/extract_hamer.py <rel_path_to_data> 
  • sample data collect for paper can be downloaded via data/download_data.sh (TODO: upload data and update script)
  1. Optional but recommended Inspect the extracted keypoints: (in osmo env)
conda activate osmo
python data_collect/glove/scripts/plot_keypoints_with_osmo.py
  1. Retarget the hamer keypoints to Psyonic to construct the dataset: (in osmo_kinematics env).

    Requires camera extrinsics represented as a (4,4) rigid transformation matrix to be saved as a .npy file and the path of the file to be provided in labs/glove2robot/config/config_extract_hamer.yaml under camera_calibration

conda activate osmo_kinematics
python data_collect/glove/kinematics/construct_retarget_dataset.py

Setting up environments

Osmo Env

conda env create -f conda/osmo.yml

Osmo Kinematics Env

conda env create -f conda/osmo_kinematics.yml

Extract Hamer

  • The most computation-intensive step of the pipeline, we automate processing of multiple experiments using data_collect/glove/labs/glove2robot/config/run_hardcoded_batches.sh

Plot Keypoints Osmo

  • Used for visually validating the extract hamer keypoints and generating the combined visualization (including the magnetometer readings)

Construct Retargeted Dataset

  • Retargets the extracted hamer keypoints to Psyonic + Franka kinematics to construct the training dataset

Train and Deploy Instructions

Policy Training and Deployment

BibTeX

If you find this work helpful, please consider citing:

@article{yin2025osmo,
    title={OSMO: Open-Source Tactile Glove for Human-to-Robot Skill Transfer},
    author={Jessica Yin and Haozhi Qi and Youngsun Wi and Sayantan Kundu and Mike Lambeta and William Yang and Changhao Wang and Tingfan Wu and Jitendra Malik and Tess Hellebrekers},
    journal={arXiv:2512.08920},
    year={2025}
}

About

Open source tactile glove for robotics research

Resources

Stars

84 stars

Watchers

3 watching

Forks

Releases

Packages

Contributors

Languages