This is the official implementation of the Open Teach including unity scripts for the VR application, teleoperation pipeline and demonstration collection pipeline.
Open Teach consists of two parts.
[x] Teleoperation using Meta Quest 3 and data collection over a range of robot morphologies and simulation environments.
[x] Policy training for various dexterous manipulation tasks across different robots and simulations.
Read VR specific information, User Interface and APK files here
Install the conda environment from the yaml file in the codebase
Allegro Sim
conda env create -f env_isaac.yml
Others
conda env create -f environment.yml
This will install all the dependencies required for the server code.
After installing all the prerequisites, you can install this pipeline as a package with pip:
pip install -e .
You can test if it had installed correctly by running import openteach from the python shell.
For Simulation specific information, follow the instructions here.
For Robot controller installation, follow the instructions here
For starting the camera sensors and streaming them inside the screen in the oculus refer here
For information on running the teleoperation and data collection refer here.
For open-source code of the policies we trained on the robots refer here
For using the API we use for policy learning, use this
For adding your own robot and simulation refer here
If you use this repo in your research, please consider citing the paper as follows:
@misc{iyer2024open,
title={OPEN TEACH: A Versatile Teleoperation System for Robotic Manipulation},
author={Aadhithya Iyer and Zhuoran Peng and Yinlong Dai and Irmak Guzey and Siddhant Haldar and Soumith Chintala and Lerrel Pinto},
year={2024},
eprint={2403.07870},
archivePrefix={arXiv},
primaryClass={cs.RO}
}