{"id":41550752,"url":"https://github.com/Open-X-Humanoid/XR-1","last_synced_at":"2026-02-02T18:01:19.691Z","repository":{"id":329368056,"uuid":"1118920938","full_name":"Open-X-Humanoid/XR-1","owner":"Open-X-Humanoid","description":null,"archived":false,"fork":false,"pushed_at":"2025-12-31T10:40:30.000Z","size":5454,"stargazers_count":31,"open_issues_count":0,"forks_count":3,"subscribers_count":1,"default_branch":"main","last_synced_at":"2026-01-02T12:46:55.704Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/Open-X-Humanoid.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-12-18T13:27:22.000Z","updated_at":"2026-01-02T09:44:01.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/Open-X-Humanoid/XR-1","commit_stats":null,"previous_names":["open-x-humanoid/xr-1"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/Open-X-Humanoid/XR-1","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Open-X-Humanoid%2FXR-1","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Open-X-Humanoid%2FXR-1/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Open-X-Humanoid%2FXR-1/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Open-X-Humanoid%2FXR-1/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Open-X-Humanoid","download_url":"https://codeload.github.com/Open-X-Humanoid/XR-1/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Open-X-Humanoid%2FXR-1/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":29016663,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-02-02T16:17:30.374Z","status":"ssl_error","status_checked_at":"2026-02-02T15:58:50.469Z","response_time":58,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2026-01-24T04:00:41.030Z","updated_at":"2026-02-02T18:01:19.685Z","avatar_url":"https://github.com/Open-X-Humanoid.png","language":"Python","funding_links":[],"categories":["1. 机器人项目 | Robots"],"sub_categories":[],"readme":"\u003cdiv align=\"center\"\u003e\n\n# XR-1: Towards Versatile Vision-Language-Action Models via Learning Unified Vision-Motion Representations \n\nA versatile and scalable vision-language-action framework: XR-1 supports robust multi-task learning across diverse robot embodiments and environments.\n\n\u003e Shichao Fan\u003csup\u003e1,\\*\u003c/sup\u003e, Kun Wu\u003csup\u003e1,\\*\u003c/sup\u003e, Zhengping Che\u003csup\u003e1,\\*,\u0026dagger;\u003c/sup\u003e, Xinhua Wang\u003csup\u003e1\u003c/sup\u003e, Di Wu\u003csup\u003e1,4\u003c/sup\u003e, Fei Liao\u003csup\u003e1\u003c/sup\u003e, Ning Liu\u003csup\u003e1\u003c/sup\u003e, Yixue Zhang\u003csup\u003e1\u003c/sup\u003e, Zhen Zhao\u003csup\u003e1\u003c/sup\u003e, Zhiyuan Xu\u003csup\u003e1\u003c/sup\u003e, Meng Li\u003csup\u003e1\u003c/sup\u003e, Qingjie Liu\u003csup\u003e3\u003c/sup\u003e, Shanghang Zhang\u003csup\u003e4\u003c/sup\u003e, Min Wan\u003csup\u003e2\u003c/sup\u003e, Jian Tang\u003csup\u003e1,\u0026#9993;\u003c/sup\u003e\n\n\u003csup\u003e1\u003c/sup\u003eBeijing Innovation Center of Humanoid Robotics,\n\u003csup\u003e2\u003c/sup\u003eSchool of Mechanical Engineering and Automation, Beihang University,\n\u003csup\u003e3\u003c/sup\u003eState Key Laboratory of Virtual Reality Technology and Systems, SCSE, Beihang University,\n\u003csup\u003e4\u003c/sup\u003eState Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University\n\n\u003csup\u003e*\u003c/sup\u003eCo-first authors,\n\u003csup\u003e\u0026dagger;\u003c/sup\u003eProject leader,\n\u003csup\u003e\u0026#9993;\u003c/sup\u003eCorresponding author,\n\n[![arXiv](https://img.shields.io/badge/arXiv-2511.02776-b31b1b.svg?logo=arxiv\u0026logoColor=white)](https://arxiv.org/abs/2511.02776) [![Project Page](https://img.shields.io/badge/Project-Page-green.svg?logo=github\u0026logoColor=white)](https://xr-1-vla.github.io/)\n\n[\\[📖 Document\\]](#documents) [\\[🚀 Quick Start\\]](#-quick-start) [\\[🤗 Models\\]](#-models) [\\[🤖 Deployment\\]](#-XR-1-Deployment) [\\[✅ Performance\\]](#-performance-in-real-world) [\\[🙋 FAQs\\]](#-faqs)\n\n\n![perform](assets/images/xr1_teaser.png)\n\n\u003c/div\u003e\n\n## TODO List\n\n- [x] Release pre-training / fine-tuning code for XR-1 series.\n- [x] Release pre-trained model, and heterogeneous dataset sample of XR-1 on both HuggingFace and ModelScope.\n- [ ] Release real world deloyment sample of XR-1.\n\n## Model Download\n- [huggingface](https://huggingface.co/collections/X-Humanoid/xr-1)\n- [modelscope](https://modelscope.cn/collections/X-Humanoid/XR-1)\n\n## Documents\nThis repository is built upon a fork of [Lerobot](https://github.com/huggingface/lerobot). Please note that due to the rapid updates of Lerobot, our implementation is specifically aligned with **Lerobot dataset v2.1**. We have preserved the original directory structure to facilitate further development and integration for the community.\n\n\n### 🚀 Quick Start\n### 📑 Installation\n\n\nDownload our source code:\n```bash\ngit clone https://github.com/Open-X-Humanoid/XR-1.git\ncd XR-1\n```\nCreate a virtual environment with Python 3.10 and activate it, e.g. with [`miniconda`](https://docs.anaconda.com/free/miniconda/index.html), then install the dependencies:\n```bash\nconda create -y -n xr1 python=3.10\nconda activate xr1\npip install -e \".[xr1]\"\n```\n\n### 📑 Dataset Preparation \n\n1.  **Format Compatibility**: Since our environment relies on **LeRobot Dataset v2.1**, we recommend using [any4lerobot](https://github.com/Tavish9/any4lerobot/) to convert your data to this standard.\n2.  **Sample Data**: We provide a heterogeneous dataset sample (including EGO4D and Robot data like TienKung2/UR/Franka) available at [X-Humanoid/XR-1-Dataset-Sample](https://huggingface.co/datasets/X-Humanoid/XR-1-Dataset-Sample). You can download it using the provided script: `scripts/hf_xr1_dataset_sample_download.sh` or `scripts/modelscope_xr1_dataset_sample_download.sh`.\n3.  **Unified Dataloader**: We have designed a powerful dataloader that unifies heterogeneous data sources and embodiments, making **pre-training** extremely simple. You can find the implementation in `examples/xr1_cross_dataset_and_embodiment_dataloader.py`.\n\n    **Key enhancements over the original LeRobot dataloader:**\n    - **Unified Data Loading**: Seamlessly reads data from diverse sources and embodiments.\n    - **Multi-Task Support**: Compatible with heterogeneous multi-task learning.\n    - **Few-Shot Capabilities**: Supports training with small sample sizes.\n    - **Extensibility**: Easily adaptable to new formats (e.g., non-LeRobot formats like Ego4D) with minimal development.\n\n## 🤗 Models\n\nTo set up the model environment, first download the foundation models (e.g., SigLIP, PaliGemma) by running:\n```bash\n# Huggingface\nbash scripts/hf_download.sh\n```\nThen, to obtain the **XR-1-Stage1-UVMC** and **XR-1-Stage2-Pretrain** models for fine-tuning, run:\n```bash\n# Huggingface\nbash scripts/hf_xr1_pretrain_model_download.sh\n# Or ModelScope\nbash modelscope_xr1_pretrain_model_download.sh\n```\n\n## 📖 Training Recipe\n\nWe provide three training paths depending on your data and performance requirements:\n\n\n### 📑 Fast Fine-tuning (For Quick Deployment)\nIf you need to quickly adapt the model to a new task or robot, you can fine-tune only the **Stage 3**. \nThis is the fastest way to get a deployable model:\n\n```bash\n# Debug Mode (For testing configurations):\nbash scripts/xr1_stage3_finetune.sh --debug\n# Standard Training (Default):\nbash scripts/xr1_stage3_finetune.sh\n\n```\n\n\n### 📑 Full Fine-tuning (Recommended for Best Performance)\nFor custom datasets where you aim for optimal performance, we strongly recommend fine-tuning **all three stages** (Stage 1, 2, and 3) sequentially to better align the representations with your specific data:\n```bash\n# Full Fine-tuning Stage1 \u0026 Stage2 \u0026 stage3\nbash scripts/xr1_stage1_finetune.sh\nbash scripts/xr1_stage2_finetune.sh\nbash scripts/xr1_stage3_finetune.sh (optional)\n```\n\n### 📑 Pre-training from Scratch\nOur framework fully supports pre-training if you have access to large-scale, heterogeneous datasets across diverse embodiments and environments:\n```bash\n# Pre-training Stage1 \u0026 Stage2\nbash scripts/xr1_stage1_pretrain.sh\nbash scripts/xr1_stage2_pretrain.sh\n```\n\n## 🤖 XR-1 Deployment\n\nWe provide a streamlined workflow to deploy and verify XR-1 on various robotic platforms, including Franka, UR, and Agilex. The following example demonstrates the process using a Franka dual-arm robot:\n``` bash\n# 1. Perform Fast Fine-tuning to train a specific Stage 3 model\n# Franka\nbash scripts/xr1_stage3_finetune.sh --debug --dataset XR_1_DATASET_DUAL_ARM_FRANKA\n# Or Tienkung2\nbash scripts/xr1_stage3_finetune.sh --debug --dataset XR_1_DATASET_DUAL_ARM_TIEN_KUNG2\n# 2. Execute the deployment script\npython deploy/real_robot/xr1_deploy.py\n```\n\nFor deployment on **TienKung 2.0**, we recommend referring to the [x-humanoid-training-toolchain](https://github.com/Open-X-Humanoid/x-humanoid-training-toolchain/tree/main/deployment) for specialized instructions.\n\n\n## ✅ Performance in Real-world\n\u003ctable width=\"100%\"\u003e\n  \u003ctr\u003e\n    \u003ctd align=\"center\" width=\"50%\"\u003e\n      \u003cimg src=\"assets/images/Dual-Arm%20UR-5e_results.png\" alt=\"Dual-Arm UR-5e\"\u003e\n      \u003cbr\u003e\u003cb\u003eDual-Arm UR-5e\u003c/b\u003e\n    \u003c/td\u003e\n    \u003ctd align=\"center\" width=\"50%\"\u003e\n      \u003cimg src=\"assets/images/Tien%20Kung%202.0_results.png\" alt=\"Tien Kung 2.0\"\u003e\n      \u003cbr\u003e\u003cb\u003eTien Kung 2.0\u003c/b\u003e\n    \u003c/td\u003e\n  \u003c/tr\u003e\n  \u003ctr\u003e\n    \u003ctd align=\"center\" width=\"50%\"\u003e\n      \u003cimg src=\"assets/images/tienkung1_results.png\" alt=\"Tien Kung 1.0\"\u003e\n      \u003cbr\u003e\u003cb\u003eTien Kung 1.0\u003c/b\u003e\n    \u003c/td\u003e\n    \u003ctd align=\"center\" width=\"50%\"\u003e\n      \u003cimg src=\"assets/images/Dual-Arm%20Franka_results.png\" alt=\"Dual-Arm Franka\"\u003e\n      \u003cbr\u003e\u003cb\u003eDual-Arm Franka\u003c/b\u003e\n    \u003c/td\u003e\n  \u003c/tr\u003e\n  \u003ctr\u003e\n    \u003ctd align=\"center\" width=\"50%\"\u003e\n      \u003cimg src=\"assets/images/AgileX%20Cobot%20Magic%20V2.0_results.png\" alt=\"AgileX Cobot Magic V2.0\"\u003e\n      \u003cbr\u003e\u003cb\u003eAgileX Cobot Magic V2.0\u003c/b\u003e\n    \u003c/td\u003e\n    \u003ctd align=\"center\" width=\"50%\"\u003e\n      \u003cimg src=\"assets/images/Single-Arm%20UR-5e_results.png\" alt=\"Single-Arm UR-5e\"\u003e\n      \u003cbr\u003e\u003cb\u003eSingle-Arm UR-5e\u003c/b\u003e\n    \u003c/td\u003e\n  \u003c/tr\u003e\n\u003c/table\u003e\n\n## 🤗 FAQs\nIf you encounter any issues, feel free to open an issue on GitHub or reach out through discussions. We appreciate your feedback and contributions! 🚀\n\n## License\n\nThis project is released under the [Apache License](LICENSE). Parts of this project contain code and models from other sources, which are subject to their respective licenses.\n\n## Citation\n\nIf you find this project useful in your research, please consider cite:\n\n```BibTeX\n@article{fan2025xr,\n  title={XR-1: Towards Versatile Vision-Language-Action Models via Learning Unified Vision-Motion Representations},\n  author={Fan, Shichao and Wu, Kun and Che, Zhengping and Wang, Xinhua and Wu, Di and Liao, Fei and Liu, Ning and Zhang, Yixue and Zhao, Zhen and Xu, Zhiyuan and others},\n  journal={arXiv preprint arXiv:2511.02776},\n  year={2025}\n}\n```\n\n## Acknowledgement\nXR-1 is built with reference to the code of the following projects: [Lerobot](https://github.com/huggingface/lerobot), [Moto](https://github.com/TencentARC/Moto), [QueST](https://github.com/pairlab/QueST) and [Pi0](https://github.com/Physical-Intelligence/openpi). Thanks for their awesome work!\n\n## Discussions\nIf you're interested in XR-1, welcome to join our WeChat group for discussions.\n\n\u003cimg src=\"./assets/images/qrocde.png\" border=0 width=30%\u003e\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FOpen-X-Humanoid%2FXR-1","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FOpen-X-Humanoid%2FXR-1","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FOpen-X-Humanoid%2FXR-1/lists"}