IsaacLab-Arena Unitree G1 Humanoid Apple Pick-and-Place: 200 XR-Teleoperated Simulation Demonstrations (NVIDIA)

Provider
NVIDIA Corporation
License
CC-BY-4.0 (source)
Size
426 files
Format
HDF5, Parquet (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/nvidia/Arena-G1-Static-PickNPlace-Task)
Access
Free and public. No account with the data provider is needed.
Category
Robotics
Source
https://huggingface.co/datasets/nvidia/Arena-G1-Static-PickNPlace-Task

About this dataset

Simulated humanoid manipulation data from NVIDIA in which a Unitree G1 humanoid robot, standing in place, picks up an apple and places it, recorded in the IsaacLab-Arena environment and used to post-train the GR00T N1.7 vision-language-action model. All 200 demonstrations were collected by a human operator with an XR (extended reality) headset teleoperating the robot in Isaac Lab at 50 Hz. The repository has the recorded 200-demo HDF5 file and a GR00T-LeRobot conversion in lerobot/ with one Parquet file and one MP4 video per episode: 43-DoF whole-body joint-position actions and joint-state observations, timestamps, language task indices, and 640x480 egocentric RGB video, plus episodes.jsonl, tasks.jsonl, modality.json and info.json. About 10 GB (measured 2026-11-21). Read it with h5py, pandas or the lerobot library. Useful for loco-manipulation imitation learning and sim-to-real comparisons with the companion real-robot AppleToPlate set. NVIDIA releases it under CC-BY-4.0 and states it is ready for commercial use; cite NVIDIA IsaacLab-Arena and GR00T N1.

How to load this dataset in Python

from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="nvidia/Arena-G1-Static-PickNPlace-Task", filename="arena_g1_static_apple_dataset_recorded_200_demos.hdf5", repo_type="dataset")
print(path)

Data structure

File types in sampled files: HDF5, Parquet

Files

Example file: arena_g1_static_apple_dataset_recorded_200_demos.hdf5

Analyze this dataset without downloading it

On Scigantic, IsaacLab-Arena Unitree G1 Humanoid Apple Pick-and-Place: 200 XR-Teleoperated Simulation Demonstrations (NVIDIA) 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