G1_BlockStacking_Dataset

This dataset was created using [LeRobot](https://github.com/huggingface/lerobot).

## Dataset Description

- **Homepage:** [More Information Needed]
- **Paper:** [More Information Needed]
- **License:** apache-2.0

- **Task Objective:** Stack the three cubic blocks on the desktop from bottom to top in the order of red, yellow, and blue on the black tape affixed to the desktop.
- **Operational Objects:** Three cubic blocks with 5cm side lengths（Purchase link:[【淘宝】假一赔四 https://e.tb.cn/h.60bRpk3V7bDhevw?tk=RzFAeJt4Wwp CZ007 「小块正方体形积木数学教具婴儿童益智拼装玩具层层叠1一2岁3宝宝6」
点击链接直接打开 或者 淘宝搜索直接打开]）.
- **Operation Duration:** Each operation takes approximately 20 to 40 seconds.
- **Recording Frequency:** 30 Hz.
- **Robot Type:** 7-DOF dual-arm G1 robot.
- **End Effector:** Three-fingered dexterous hands.
- **Dual-Arm Operation:** Yes.
- **Image Resolution:** 640x480.
- **Camera Positions:** Wrist-mounted (monocular camera) + head-mounted (binocular cameras).
- **Data Content:**
• Robot's current state.
• Robot's next action.
• Current camera view images.
- **Robot Initial Posture:** The first robot state in each dataset entry.
- **Object Placement:** Randomly placed within the robot arm's motion range and the field of view of the robot's head-mounted camera.
- **Camera View:** Follow the guidelines in **Part 5** of [AVP Teleoperation Documentation](https://github.com/unitreerobotics/avp_teleoperate).

- **Important Notes:**
1. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in **Part 5** of [AVP Teleoperation Documentation](https://github.com/unitreerobotics/avp_teleoperate).
2. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training.

## Dataset Structure

[meta/info.json](meta/info.json):
```json
{
"codebase_version": "v2.0",
"robot_type": "Unitree_G1",
"total_episodes": 301,
"total_frames": 281196,
"total_tasks": 1,
"total_videos": 1204,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:301"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path": "videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4",
"features": {
"observation.state": {
"dtype": "float32",
"shape": [
28
],
"names": [
[
"kLeftShoulderPitch",
"kLeftShoulderRoll",
"kLeftShoulderYaw",
"kLeftElbow",
"kLeftWristRoll",
"kLeftWristPitch",
"kLeftWristYaw",
"kRightShoulderPitch",
"kRightShoulderRoll",
"kRightShoulderYaw",
"kRightElbow",
"kRightWristRoll",
"kRightWristPitch",
"kRightWristYaw",
"kLeftHandThumb0",
"kLeftHandThumb1",
"kLeftHandThumb2",
"kLeftHandMiddle0",
"kLeftHandMiddle1",
"kLeftHandIndex0",
"kLeftHandIndex1",
"kRightHandThumb0",
"kRightHandThumb1",
"kRightHandThumb2",
"kRightHandIndex0",
"kRightHandIndex1",
"kRightHandMiddle0",
"kRightHandMiddle1"
]
]
},
"action": {
"dtype": "float32",
"shape": [
28
],
"names": [
[
"kLeftShoulderPitch",
"kLeftShoulderRoll",
"kLeftShoulderYaw",
"kLeftElbow",
"kLeftWristRoll",
"kLeftWristPitch",
"kLeftWristYaw",
"kRightShoulderPitch",
"kRightShoulderRoll",
"kRightShoulderYaw",
"kRightElbow",
"kRightWristRoll",
"kRightWristPitch",
"kRightWristYaw",
"kLeftHandThumb0",
"kLeftHandThumb1",
"kLeftHandThumb2",
"kLeftHandMiddle0",
"kLeftHandMiddle1",
"kLeftHandIndex0",
"kLeftHandIndex1",
"kRightHandThumb0",
"kRightHandThumb1",
"kRightHandThumb2",
"kRightHandIndex0",
"kRightHandIndex1",
"kRightHandMiddle0",
"kRightHandMiddle1"
]
]
},
"observation.images.cam_left_high": {
"dtype": "video",
"shape": [
3,
480,
640
],
"names": [
"channels",
"height",
"width"
],
"info": {
"video.fps": 30.0,
"video.height": 480,
"video.width": 640,
"video.channels": 3,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"has_audio": false
}
},
"observation.images.cam_right_high": {
"dtype": "video",
"shape": [
3,
480,
640
],
"names": [
"channels",
"height",
"width"
],
"info": {
"video.fps": 30.0,
"video.height": 480,
"video.width": 640,
"video.channels": 3,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"has_audio": false
}
},
"observation.images.cam_left_wrist": {
"dtype": "video",
"shape": [
3,
480,
640
],
"names": [
"channels",
"height",
"width"
],
"info": {
"video.fps": 30.0,
"video.height": 480,
"video.width": 640,
"video.channels": 3,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"has_audio": false
}
},
"observation.images.cam_right_wrist": {
"dtype": "video",
"shape": [
3,
480,
640
],
"names": [
"channels",
"height",
"width"
],
"info": {
"video.fps": 30.0,
"video.height": 480,
"video.width": 640,
"video.channels": 3,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"has_audio": false
}
},
"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
}
}
}
```

## Citation

**BibTeX:**

```bibtex
[More Information Needed]
```

本数据集基于[LeRobot](https://github.com/huggingface/lerobot)框架构建。

## 数据集说明

- **主页**：[待补充更多信息]
- **论文**：[待补充更多信息]
- **许可证**：Apache-2.0

- **任务目标**：将桌面上粘贴的黑色胶带区域内的三个立方体积木，按红、黄、蓝的顺序自下而上堆叠于桌面。
- **操作对象**：边长为5cm的三个立方体积木（购买链接：[【淘宝】假一赔四 https://e.tb.cn/h.60bRpk3V7bDhevw?tk=RzFAeJt4Wwp CZ007 「小块正方体形积木数学教具婴儿童益智拼装玩具层层叠1一2岁3宝宝6」点击链接直接打开 或 淘宝搜索直接打开]）。
- **操作时长**：单次操作耗时约20至40秒。
- **录制帧率**：30 Hz。
- **机器人类型**：7自由度双臂G1机器人。
- **末端执行器**：三指灵巧手。
- **双臂操作**：是。
- **图像分辨率**：640×480。
- **相机布局**：腕部安装单目相机 + 头部安装双目相机。
- **数据内容**：
· 机器人当前状态
· 机器人待执行动作
· 当前相机视角图像
- **机器人初始姿态**：每条数据条目对应的首个机器人状态。
- **物体摆放**：随机放置于机械臂运动范围内以及头部相机的视场中。
- **相机视角规范**：遵循[AVP远程操作文档](https://github.com/unitreerobotics/avp_teleoperate)的第5部分指南。

- **重要说明**：
1. 由于无法精准描述空间位置，需按照[AVP远程操作文档](https://github.com/unitreerobotics/avp_teleoperate)第5部分的要求完成硬件安装后，调整场景以使其与数据集的首帧画面高度匹配。
2. 数据集并非单次采集完成，不同数据条目间存在差异，在模型训练阶段需充分考虑此类差异。

## 数据集结构

[meta/info.json](meta/info.json):
json
{
"codebase_version": "v2.0",
"robot_type": "Unitree_G1",
"total_episodes": 301,
"total_frames": 281196,
"total_tasks": 1,
"total_videos": 1204,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:301"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path": "videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4",
"features": {
"observation.state": {
"dtype": "float32",
"shape": [
28
],
"names": [
[
"kLeftShoulderPitch",
"kLeftShoulderRoll",
"kLeftShoulderYaw",
"kLeftElbow",
"kLeftWristRoll",
"kLeftWristPitch",
"kLeftWristYaw",
"kRightShoulderPitch",
"kRightShoulderRoll",
"kRightShoulderYaw",
"kRightElbow",
"kRightWristRoll",
"kRightWristPitch",
"kRightWristYaw",
"kLeftHandThumb0",
"kLeftHandThumb1",
"kLeftHandThumb2",
"kLeftHandMiddle0",
"kLeftHandMiddle1",
"kLeftHandIndex0",
"kLeftHandIndex1",
"kRightHandThumb0",
"kRightHandThumb1",
"kRightHandThumb2",
"kRightHandIndex0",
"kRightHandIndex1",
"kRightHandMiddle0",
"kRightHandMiddle1"
]
]
},
"action": {
"dtype": "float32",
"shape": [
28
],
"names": [
[
"kLeftShoulderPitch",
"kLeftShoulderRoll",
"kLeftShoulderYaw",
"kLeftElbow",
"kLeftWristRoll",
"kLeftWristPitch",
"kLeftWristYaw",
"kRightShoulderPitch",
"kRightShoulderRoll",
"kRightShoulderYaw",
"kRightElbow",
"kRightWristRoll",
"kRightWristPitch",
"kRightWristYaw",
"kLeftHandThumb0",
"kLeftHandThumb1",
"kLeftHandThumb2",
"kLeftHandMiddle0",
"kLeftHandMiddle1",
"kLeftHandIndex0",
"kLeftHandIndex1",
"kRightHandThumb0",
"kRightHandThumb1",
"kRightHandThumb2",
"kRightHandIndex0",
"kRightHandIndex1",
"kRightHandMiddle0",
"kRightHandMiddle1"
]
]
},
"observation.images.cam_left_high": {
"dtype": "video",
"shape": [
3,
480,
640
],
"names": [
"channels",
"height",
"width"
],
"info": {
"video.fps": 30.0,
"video.height": 480,
"video.width": 640,
"video.channels": 3,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"has_audio": false
}
},
"observation.images.cam_right_high": {
"dtype": "video",
"shape": [
3,
480,
640
],
"names": [
"channels",
"height",
"width"
],
"info": {
"video.fps": 30.0,
"video.height": 480,
"video.width": 640,
"video.channels": 3,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"has_audio": false
}
},
"observation.images.cam_left_wrist": {
"dtype": "video",
"shape": [
3,
480,
640
],
"names": [
"channels",
"height",
"width"
],
"info": {
"video.fps": 30.0,
"video.height": 480,
"video.width": 640,
"video.channels": 3,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"has_audio": false
}
},
"observation.images.cam_right_wrist": {
"dtype": "video",
"shape": [
3,
480,
640
],
"names": [
"channels",
"height",
"width"
],
"info": {
"video.fps": 30.0,
"video.height": 480,
"video.width": 640,
"video.channels": 3,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"has_audio": false
}
},
"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
}
}
}

## 引用

**BibTeX格式**：
bibtex
[More Information Needed]

搜集汇总

背景与挑战

背景概述

G1_BlockStacking_Dataset是一个用于训练机器人完成积木堆叠任务的数据集，包含301个操作片段和281196帧数据，记录了7-DOF双臂G1机器人在堆叠三个立方体积木时的状态、动作和摄像头视图图像。数据集的特点是操作对象明确（红、黄、蓝三个立方体积木），操作时间短（20-40秒），且数据采集频率高（30Hz），适用于机器人学习和控制算法的研究。