so101-pick-block-place-in-box_20260605_175323
收藏资源简介:
该数据集由LeRobot创建,专注于机器人操作任务,具体为拾取-放置方块到盒子中(pick-block-place-in-box)。数据集包含机器人动作和观察状态数据,动作数据包括6个关节位置(如肩部平移、肩部升降、肘部弯曲等),观察状态包括相同关节位置以及来自前部和侧部摄像头的视频数据。视频数据以30fps的帧率记录,分辨率为480x640像素,使用AV1编码。数据集总共有14个episodes,12558帧,覆盖1个任务,数据以parquet文件格式组织,视频文件以mp4格式存储。适用于机器人学习、强化学习和计算机视觉研究。
This dataset is created by LeRobot and focuses on robot manipulation tasks, specifically pick-block-place-in-box. It includes robot action and observation state data. Action data consists of 6 joint positions (e.g., shoulder translation, shoulder lift, elbow bend, etc.), while observation states include the same joint positions and video data from front and side cameras. The video data is recorded at 30fps with a resolution of 480x640 pixels, using AV1 encoding. The dataset contains 14 episodes, 12558 frames, and covers 1 task. Data is organized in parquet file format, with video files stored in mp4 format. It is suitable for robot learning, reinforcement learning, and computer vision research.
数据集概述
- 名称: so101-pick-block-place-in-box_20260605_175323
- 任务类别: 机器人学 (Robotics)
- 许可证: Apache-2.0
- 创建工具: LeRobot (https://github.com/huggingface/lerobot)
数据集结构
- 总片段数 (Episodes): 14
- 总帧数 (Frames): 12,558
- 总任务数 (Tasks): 1
- 帧率 (FPS): 30
- 数据文件大小: 100 MB
- 视频文件大小: 200 MB
- 机器人类型: so_follower
数据特征
| 特征名称 | 数据类型 | 形状 | 说明 |
|---|---|---|---|
action |
float32 | [6] | 包含肩关节、肘关节、腕关节和夹爪的位置指令 |
observation.state |
float32 | [6] | 机器人6个关节的状态(位置) |
observation.images.front |
video (AV1编码) | [480, 640, 3] | 前方摄像头视频,分辨率480×640,30 FPS |
observation.images.side |
video (AV1编码) | [480, 640, 3] | 侧方摄像头视频,分辨率480×640,30 FPS |
timestamp |
float32 | [1] | 时间戳 |
frame_index |
int64 | [1] | 帧索引 |
episode_index |
int64 | [1] | 片段索引 |
index |
int64 | [1] | 全局索引 |
task_index |
int64 | [1] | 任务索引 |
数据划分
- 训练集 (train): 片段 0 到 13(所有14个片段均用于训练)
文件存储格式
- 数据文件: Parquet 格式,路径为
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频文件: MP4 格式,路径为
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4



