five

so101_placeinbin_red_20260503_164829_clean

收藏
Hugging Face2026-05-19 更新2026-05-19 收录
下载链接:
https://huggingface.co/datasets/RobotLearningProject/so101_placeinbin_red_20260503_164829_clean
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是一个专门用于机器人学习任务的数据集,专注于放置物品到垃圾桶的动作。它包含30个剧集,总计10826帧,帧率为30fps。数据集采用so_follower机器人类型,代码库版本为v3.0。数据以parquet格式存储,总大小为100MB,视频文件大小为200MB,视频分辨率为720x1280,编码为av1。特征包括动作(包含6个关节位置:肩部平移、肩部提升、肘部弯曲、手腕弯曲、手腕滚动、夹爪位置)、观察状态(同样包含6个关节位置)、观察图像(来自前端摄像头的视频,分辨率为720x1280,3通道)、时间戳、帧索引、剧集索引、索引和任务索引。数据集仅包含训练分割,覆盖所有剧集。

This dataset is tailored for robotic learning tasks, focusing on the action of placing objects into trash bins. It contains 30 episodes, totaling 10,826 frames with a frame rate of 30 fps. The dataset adopts the so_follower robot type, with a codebase version of v3.0. The data is stored in Parquet format, with a total size of 100 MB; the video files have a total size of 200 MB, with a resolution of 720×1280 and encoded in AV1. The included features are as follows: actions (comprising 6 joint positions: shoulder translation, shoulder lift, elbow flexion, wrist flexion, wrist roll, and gripper position), observation states (also containing 6 joint positions), observation images (videos captured by the front-facing camera, with a resolution of 720×1280 and 3 channels), timestamps, frame indices, episode indices, index, and task indices. The dataset only includes a training split covering all episodes.
提供机构:
RobotLearningProject
创建时间:
2026-05-19
原始信息汇总

数据集概述

  • 数据集名称:so101_placeinbin_red_20260503_164829_clean
  • 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)

数据集基本信息

  • 机器人类型:so_follower
  • 总片段数:30
  • 总帧数:10,826
  • 总任务数:1
  • 数据块大小:1,000
  • FPS:30
  • 数据文件大小:约100 MB
  • 视频文件大小:约200 MB
  • 数据集划分:训练集(0:30)

数据结构

数据文件路径

  • 数据文件:data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
  • 视频文件:videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4

特征(Features)

特征名 数据类型 形状 说明
action float32 (6,) 包含6个关节的动作:shoulder_pan.pos、shoulder_lift.pos、elbow_flex.pos、wrist_flex.pos、wrist_roll.pos、gripper.pos
observation.state float32 (6,) 与action相同,包含6个关节的状态
observation.images.front video (720, 1280, 3) 前置摄像头视频,分辨率720x1280,3通道,编码为av1,帧率30fps
timestamp float32 (1,) 时间戳
frame_index int64 (1,) 帧索引
episode_index int64 (1,) 片段索引
index int64 (1,) 索引
task_index int64 (1,) 任务索引
二维码
社区交流群
二维码
科研交流群
商业服务