Xp-Teleop: Long-Horizon Humanoid Loco-Manipulation Benchmark (Unitree G1)
- Provider
- Ropedia
- License
- CC-BY-4.0 (source)
- Size
- 4,354 files
- Format
- JSON Lines, JSON (in sampled files)
- Stored at
- Hugging Face (https://huggingface.co/datasets/ropedia-ai/Xp-Teleop)
- Access
- Free and public. No account with the data provider is needed.
- Category
- Robotics
- Source
- https://arxiv.org/abs/2610.26855
About Xp-Teleop: Long-Horizon Humanoid Loco-Manipulation Benchmark (Unitree G1)
Xp-Teleop is the long-horizon humanoid loco-manipulation benchmark introduced with Ropedia W0 (arXiv:2610.26855). It records a Unitree G1 humanoid with Inspire FTP hands from a head-mounted RGB camera and cuts each teleoperated episode into individual actions. The card reports 27 episodes, 543 action segments and 96,064 frames at 30 fps and 1280x720. Each segment folder under takes/ holds clean RGB video (H.264 MP4), the same video with skeleton overlays, original frame timestamps (CSV), per-frame arrays (frames.npz), robot body masks (PNG in a zip), head IMU at about 1.6 kHz (CSV), Inspire tactile readings with 1,062 taxels per hand (JSONL) and the action caption (JSON). meta/ holds the schema (info.json), takes, segments, tasks and per-episode camera calibration as JSONL. Measured 2026-11-14 at about 7.9 GB. Read with numpy, pandas and PyAV or OpenCV. Useful for humanoid world models, action segmentation, video-language grounding and tactile-aware manipulation. Licensed under CC BY 4.0; cite the Ropedia W0 paper.
How to load this dataset in Python
from datasets import load_dataset
ds = load_dataset("ropedia-ai/Xp-Teleop")
print(ds)Data structure
File types in sampled files: JSON Lines, JSON
Files
Example file: meta/calibration.jsonl
Analyze this dataset without downloading it
On Scigantic, Xp-Teleop: Long-Horizon Humanoid Loco-Manipulation Benchmark (Unitree G1) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.
This page is also available as Markdown and as JSON.
Related Robotics datasets
- OctoSense: Synchronized Multi-Sensor Driving Dataset with Event Cameras
- Open X-Embodiment
- Ouster OS1-64 LiDAR Drives with Aptiv ESR Radar Tracks, Kalamazoo 2020-2025 (WMU AD Assurance Lab)
- Infrastructure-Enabled Energy-Efficient Autonomous Vehicles (DOE EEMS116): Radar Retroreflector, Smart Pavement Marker, HD-Map and Weather Sensing Recordings
- X-VLA Soft Cloth Folding Dataset
- PhysicalAI Robotics Manipulation SingleArm
- DexCap: Hand Motion Capture Data for Dexterous Manipulation
- NuRec AV Object Benchmark