bi_so101_flatten-and-fold-the-rag-then-place-0517_20260517_202324
收藏资源简介:
该数据集是一个机器人操作任务数据集,使用LeRobot工具创建。数据集包含一个名为bi_so_follower的双臂机器人的操作数据,涉及展平和折叠抹布然后放置的任务。数据特征包括:动作(action)数据,包含12个浮点数值,分别对应左右机械臂的肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置;观测(observation)数据,包含机器人状态(同样12维浮点数)和来自三个摄像头的图像数据(left_camera1、left_camera3和right_camera2),图像分辨率为480x640,视频编码为h264,帧率为30fps。数据集还包含时间戳、帧索引、回合索引、任务索引等元数据。数据集总共有1个任务、1个回合、7750帧数据,数据文件总大小为100MB,视频文件总大小为200MB。数据集以Parquet格式存储,视频以MP4格式存储。
This dataset is a robotic manipulation task dataset created using the LeRobot toolkit. It contains manipulation data from a dual-arm robot named bi_so_follower, covering the tasks of flattening, folding, and placing rags. The dataset includes two core data modalities: action data and observation data. The action data consists of 12 floating-point values, which respectively correspond to the shoulder translation, shoulder lift, elbow flexion, wrist flexion, wrist rotation, and gripper position of the left and right robotic arms. The observation data includes robot state (also a 12-dimensional floating-point vector) and image data captured by three cameras: left_camera1, left_camera3, and right_camera2. The images have a resolution of 480×640, use H.264 video encoding, and operate at a frame rate of 30 fps. Additional metadata fields include timestamps, frame indices, episode indices, and task indices. In total, the dataset contains 1 task, 1 episode, and 7750 frames of data. The combined size of the data files is 100 MB, while the total size of the video files is 200 MB. The dataset is stored in Parquet format, and the video files are stored in MP4 format.
数据集概述
- 数据集名称:bi_so101_flatten-and-fold-the-rag-then-place-0517_20260517_202324
- 数据集地址:https://huggingface.co/datasets/wuc1/bi_so101_flatten-and-fold-the-rag-then-place-0517_20260517_202324
- 许可证:Apache-2.0
- 任务类别:机器人学(robotics)
- 标签:LeRobot
数据集结构
该数据集使用 LeRobot 创建,主要结构如下:
- 数据格式:Parquet 文件(
data/*/*.parquet) - 机器人类型:
bi_so_follower - 总集数(episodes):1
- 总帧数(frames):7750
- 总任务数(tasks):1
- 数据分块:数据以
chunk-{chunk_index:03d}/file-{file_index:03d}.parquet组织,视频以videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4组织 - 数据大小:数据文件约 100 MB,视频文件约 200 MB
- 帧率:30 fps
数据集特征
| 特征名称 | 数据类型 | 形状 | 说明 |
|---|---|---|---|
action |
float32 | [12] | 包含左右双臂的关节角度和夹爪位置(共12个维度) |
observation.state |
float32 | [12] | 与 action 相同的状态观测(12维) |
observation.images.left_camera1 |
video | [480, 640, 3] | 左侧摄像头1视频(h264编码,yuv420p,30fps) |
observation.images.left_camera3 |
video | [480, 640, 3] | 左侧摄像头3视频(h264编码,yuv420p,30fps) |
observation.images.right_camera2 |
video | [480, 640, 3] | 右侧摄像头2视频(h264编码,yuv420p,30fps) |
timestamp |
float32 | [1] | 时间戳 |
frame_index |
int64 | [1] | 帧索引 |
episode_index |
int64 | [1] | 集索引 |
index |
int64 | [1] | 全局索引 |
task_index |
int64 | [1] | 任务索引 |
数据集划分
- 训练集:
train划分包含所有数据(索引 0:1)
引用信息
该数据集尚未提供正式的引用格式(BibTeX 信息为 [More Information Needed])。



