HIW-500-LeRobot / README.md
aarontung's picture
Update README.md
d935d49 verified
|
Raw
History Blame Contribute Delete
10.9 kB
metadata
pretty_name: 'HIW-500: Humanoids In-the-Wild Dataset (LeRobot)'
language:
  - en
tags:
  - LeRobot
  - robotics
  - humanoid
  - unitree-g1
  - manipulation
  - mobile-manipulation
  - bimanual
  - imitation-learning
  - teleoperation
  - in-the-wild
task_categories:
  - robotics
configs:
  - config_name: default
    data_files: data/*/*.parquet
license: cc-by-4.0

This dataset was created using LeRobot.

Dataset Structure

meta/info.json:

{
    "codebase_version": "v3.0",
    "fps": 30,
    "features": {
        "observation.images.head": {
            "dtype": "video",
            "shape": [
                480,
                1280,
                3
            ],
            "names": [
                "height",
                "width",
                "channels"
            ],
            "info": {
                "video.height": 480,
                "video.width": 1280,
                "video.codec": "av1",
                "video.pix_fmt": "yuv420p",
                "video.is_depth_map": false,
                "video.fps": 30,
                "video.channels": 3,
                "has_audio": false,
                "video.g": 2,
                "video.crf": 30,
                "video.preset": 12,
                "video.fast_decode": 0,
                "video.video_backend": "pyav",
                "video.extra_options": {}
            }
        },
        "observation.images.left_wrist": {
            "dtype": "video",
            "shape": [
                480,
                640,
                3
            ],
            "names": [
                "height",
                "width",
                "channels"
            ],
            "info": {
                "video.height": 480,
                "video.width": 640,
                "video.codec": "av1",
                "video.pix_fmt": "yuv420p",
                "video.is_depth_map": false,
                "video.fps": 30,
                "video.channels": 3,
                "has_audio": false,
                "video.g": 2,
                "video.crf": 30,
                "video.preset": 12,
                "video.fast_decode": 0,
                "video.video_backend": "pyav",
                "video.extra_options": {}
            }
        },
        "observation.images.right_wrist": {
            "dtype": "video",
            "shape": [
                480,
                640,
                3
            ],
            "names": [
                "height",
                "width",
                "channels"
            ],
            "info": {
                "video.height": 480,
                "video.width": 640,
                "video.codec": "av1",
                "video.pix_fmt": "yuv420p",
                "video.is_depth_map": false,
                "video.fps": 30,
                "video.channels": 3,
                "has_audio": false,
                "video.g": 2,
                "video.crf": 30,
                "video.preset": 12,
                "video.fast_decode": 0,
                "video.video_backend": "pyav",
                "video.extra_options": {}
            }
        },
        "observation.state": {
            "dtype": "float32",
            "shape": [
                29
            ],
            "names": [
                "kLeftHipPitch.q",
                "kLeftHipRoll.q",
                "kLeftHipYaw.q",
                "kLeftKnee.q",
                "kLeftAnklePitch.q",
                "kLeftAnkleRoll.q",
                "kRightHipPitch.q",
                "kRightHipRoll.q",
                "kRightHipYaw.q",
                "kRightKnee.q",
                "kRightAnklePitch.q",
                "kRightAnkleRoll.q",
                "kWaistYaw.q",
                "kWaistRoll.q",
                "kWaistPitch.q",
                "kLeftShoulderPitch.q",
                "kLeftShoulderRoll.q",
                "kLeftShoulderYaw.q",
                "kLeftElbow.q",
                "kLeftWristRoll.q",
                "kLeftWristPitch.q",
                "kLeftWristyaw.q",
                "kRightShoulderPitch.q",
                "kRightShoulderRoll.q",
                "kRightShoulderYaw.q",
                "kRightElbow.q",
                "kRightWristRoll.q",
                "kRightWristPitch.q",
                "kRightWristYaw.q"
            ]
        },
        "observation.state.wbc": {
            "dtype": "float32",
            "shape": [
                23
            ],
            "names": [
                "pivot_vx",
                "pivot_vy",
                "pivot_vyaw",
                "pivot_roll",
                "pivot_pitch",
                "pivot_yaw",
                "pivot_height",
                "left_ee_x",
                "left_ee_y",
                "left_ee_z",
                "left_ee_roll",
                "left_ee_pitch",
                "left_ee_yaw",
                "right_ee_x",
                "right_ee_y",
                "right_ee_z",
                "right_ee_roll",
                "right_ee_pitch",
                "right_ee_yaw",
                "left_trigger",
                "left_squeeze",
                "right_trigger",
                "right_squeeze"
            ]
        },
        "action": {
            "dtype": "float32",
            "shape": [
                23
            ],
            "names": [
                "pivot_vx",
                "pivot_vy",
                "pivot_vyaw",
                "pivot_roll",
                "pivot_pitch",
                "pivot_yaw",
                "pivot_height",
                "left_ee_x",
                "left_ee_y",
                "left_ee_z",
                "left_ee_roll",
                "left_ee_pitch",
                "left_ee_yaw",
                "right_ee_x",
                "right_ee_y",
                "right_ee_z",
                "right_ee_roll",
                "right_ee_pitch",
                "right_ee_yaw",
                "left_trigger",
                "left_squeeze",
                "right_trigger",
                "right_squeeze"
            ]
        },
        "timestamp": {
            "dtype": "float32",
            "shape": [
                1
            ],
            "names": null
        },
        "frame_index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "episode_index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "task_index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "language_persistent": {
            "dtype": "language",
            "shape": [
                1
            ],
            "names": null
        },
        "language_events": {
            "dtype": "language",
            "shape": [
                1
            ],
            "names": null
        }
    },
    "total_episodes": 23743,
    "total_frames": 40839947,
    "total_tasks": 11,
    "chunks_size": 1000,
    "data_files_size_in_mb": 100,
    "video_files_size_in_mb": 200,
    "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
    "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4",
    "robot_type": "unitree_g1",
    "splits": {
        "train": "0:23743"
    }
}

HIW-500: Humanoids In-the-Wild Dataset (LeRobot format)

https://bitrobot-foundation.github.io/humanoids-in-the-wild-500-hours/

This is the LeRobot v3.0 conversion of BitRobot/HIW-500. HIW-500 is a large-scale dataset for whole-body humanoid robot learning in natural home environments. It captures human teleoperation demonstrations on Unitree G1 across real homes in Southeast Asia, where layouts, object states, lighting, clutter, and operator styles vary from episode to episode.

The dataset is designed for research on mobile manipulation, bimanual interaction, long-horizon household skills, imitation learning, and general-purpose robot learning from in-the-wild demonstrations.

Dataset Overview

  • 500+ hours of humanoid robot demonstrations
  • 23K+ episodes
  • Around 10 TB of data
  • 10+ household tasks
  • 12 real homes
  • 161 subtask labels
  • 148K+ subtask annotations

Data Modalities

Each episode records human whole-body teleoperation of Unitree G1 in real homes. The dataset combines synchronized visual observations, robot states, actions, and metadata.

Camera Streams

  • Head camera: RGB stereo, 480p, 30 FPS
  • Wrist camera: RGB, stereo IR, 480p, 30 FPS

Robot State and Actions

  • 29-DoF joint states
  • End-effector state
  • IMU
  • Odometry
  • Action traces from human whole-body teleoperation

Metadata

  • Language annotations
  • Episode information
  • Camera intrinsics and extrinsics

Dataset Statistics

Task duration

Task episodes

Average duration

Dataset Access

The dataset is hosted on Hugging Face in two formats:

License

The public HIW-500 dataset is released under the CC BY 4.0 license. If you're interested in additional datasets similar to HIW-500, be it for commercial or academic purposes, pls contact us.

Citation

If you use this dataset, please cite:

@misc{hiw500_2026,
  title={HIW-500: Humanoids In-the-Wild Dataset for Robot Learning},
  author={BitRobot and Unitree and Hugging Face},
  year={2026},
  howpublished={\url{https://bitrobot-foundation.github.io/humanoids-in-the-wild-500-hours/}}
}

Commercial Access and Custom Data

For additional coverage, commercial rights, evaluation data, or custom data collection, contact us.