Skip to content

nota-github/xpu-leisaac

 
 

Repository files navigation

LeIsaac 🚀

leisaac.mp4

This repository provides teleoperation functionality in IsaacLab using the SO101Leader (LeRobot), including data collection, data conversion, and subsequent policy training.

  • 🤖 We use SO101Follower as the robot (and other related robot) in IsaacLab and provide relevant teleoperation method.
  • 🔄 We offer scripts to convert data from HDF5 format to the LeRobot Dataset.
  • 🧠 We utilize simulation-collected data to fine-tune GR00T N1.5 and deploy it on real hardware. And more policies will be supported.

Remote Policy Evaluation

LeIsaac can also run rendering-side evaluation while the policy is hosted on a separate remote inference server. In this setup:

  • The rendering server runs Isaac Sim, the environment, cameras, and scripts/evaluation/policy_inference.py.
  • The policy inference server hosts the model and answers policy requests over ZMQ.
  • The rendering server sends observations to the policy server, receives actions, and executes the rollout in simulation.

Typical usage:

python scripts/evaluation/policy_inference.py \
  --headless \
  --enable_cameras \
  --device=cuda:0 \
  --task=LeIsaac-SO101-PickOrange-v0 \
  --eval_rounds=10 \
  --policy_type=molmoact2_so101 \
  --policy_host=<policy-server-host> \
  --policy_port=5556 \
  --policy_timeout_ms=30000 \
  --policy_action_horizon=30 \
  --policy_language_instruction="Pick three oranges and put them into the plate, then reset the arm to rest state."

Use --policy_port=5555 for GR00T N1.5 and --policy_port=5556 for MolmoAct2 SO101 compatibility. For multi-host deployments, make sure the rendering server can reach the policy server over the configured TCP port.

Tip

Welcome to the Lightwheel open-source community!

Join us, contribute, and help shape the future of AI and robotics. For questions or collaboration, contact Zeyu or Yinghao.

News 🗞️

  • [26/04/14] Remote teleoperation is now available in LeIsaac! Try it out here.
  • [26/03/10] With the new datagen module, LeIsaac can generate motion trajectories programmatically. See State Machine Data Generation.
  • [26/01/16] Added inference support for GR00T N1.6; details are in Available Policy Inference.
  • [26/01/13] Try our tutorial LeIsaac x Cosmos to get a video2action data generation pipeline.
  • [26/01/12] Extra feature of lerobot recorder integration released! You can now record data directly in LeRobot Dataset format during teleoperation.
  • [25/12/19] Try our tutorial LeIsaac x Marble to build and evaluate diverse embodied tasks across large-scale generalized environments.
  • [25/12/19] We now support lekiwi-based teleoperation and provide a Loft scene for the community. See the example task here.
  • [25/11/27] We now support more teleoperation devices, including the enhanced keyboard and gamepad. Refer to the device guide for usage details.
  • [25/11/26] LeIsaac is now the official imitation-learning (IL) simulation playground integrated into LeRobot’s EnvHub. It provides fast, scalable Isaac-based environments designed for imitation learning, control, and policy evaluation. It is featured in LeRobot: LeIsaac × LeRobot EnvHub

Getting Started 📚

Please refer to our documentation to learn how to use this repository. Follow these links to learn more about:

Contributing 🤝

Please see CONTRIBUTING.md for how to report issues and submit pull requests.

Citation 📝

If you use leisaac, please cite it as follows.

@software{Lightwheel_and_LeIsaac_Project_Developers_LeIsaac_2025,
author = {{Lightwheel} and {LeIsaac Project Developers}},
license = {Apache-2.0},
month = dec,
title = {{LeIsaac}},
url = {https://github.com/LightwheelAI/leisaac},
version = {0.4.0},
year = {2026}
}

Acknowledgements 🙏

We gratefully acknowledge IsaacLab and LeRobot for their excellent work, from which we have borrowed some code.

Join Our Team! 💼

We're always looking for talented individuals passionate about AI and robotics! If you're interested in:

  • 🤖 Robotics Engineering: Working with cutting-edge robotic systems and teleoperation
  • 🧠 AI/ML Research: Developing next-generation AI models for robotics
  • 💻 Software Engineering: Building robust, scalable robotics software
  • 🔬 Research & Development: Pushing the boundaries of what's possible in robotics

Join us at Lightwheel AI! We offer:

  • Competitive compensation and benefits
  • Work with state-of-the-art robotics technology
  • Collaborative, innovative environment
  • Opportunity to shape the future of AI-powered robotics

Apply Now → | Contact Now → | Learn More About Us →


Let's build the future of robotics together! 🤝


About

LeIsaac provides teleoperation functionality in IsaacLab using the SO101Leader (LeRobot), including data collection, data conversion, and subsequent policy training.

Resources

License

Contributing

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • Python 100.0%