遇见数据集

bimanual-so101-pickvials-real-10fps

收藏
Hugging Face2026-08-12 更新2026-07-22 收录
官方服务:

资源简介:

该数据集是一个用于机器人学任务的真实世界双手操作数据集,专注于拾取小瓶(pickvials)的场景。它包含50个训练片段,总帧数为22405,帧率为10fps。数据特征包括:动作数据(12维浮点数,表示左右机械臂的肩部、肘部、腕部关节位置以及夹爪位置);观察状态数据(同样12维浮点数,反映机械臂的实时状态);视频观察数据,来自三个摄像头视角(左腕、中心和右腕),每个视频的分辨率为480x640,3通道,使用AV1编码,帧率为10fps;以及元数据如时间戳、帧索引、片段索引、索引和任务索引。数据集以Parquet格式存储,总数据文件大小为100MB,视频文件大小为200MB,适用于机器人控制、模仿学习等研究。

This dataset is a real-world dual-arm manipulation dataset designed for robotics tasks, focusing on the scenario of small vial picking (pickvials). It contains 50 training episodes, with a total of 22405 frames and a frame rate of 10 fps. The data features include: action data, which is 12-dimensional floating-point numbers representing the shoulder, elbow, wrist joint positions and gripper positions of the left and right robotic arms; observation state data, also 12-dimensional floating-point numbers reflecting the real-time state of the robotic arms; video observation data from three camera viewpoints: left wrist, center, and right wrist, with each video having a resolution of 480×640, 3 channels, using AV1 encoding and a frame rate of 10 fps; as well as metadata such as timestamps, frame indices, episode indices, indices, and task indices. The dataset is stored in Parquet format, with a total data file size of 100 MB and a video file size of 200 MB, and is applicable to research in areas such as robotic control and imitation learning.

提供机构:
ChihHanShen
创建时间:
2026-08-12
原始信息汇总

数据集概述

  • 数据集名称: bimanual-so101-pickvials-real-10fps
  • 许可证: Apache-2.0
  • 任务类别: 机器人学(Robotics)
  • 标签: LeRobot, so101, bimanual
  • 创建工具: LeRobot
  • 机器人类型: bi_so_follower

数据集结构

  • 总片段数: 50
  • 总帧数: 22405
  • 总任务数: 1
  • 帧率: 10 FPS
  • 数据大小: 100 MB(数据文件),200 MB(视频文件)
  • 数据文件路径: data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
  • 视频文件路径: videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
  • 数据划分: 仅含训练集(train: 0:50

特征字段

动作与状态(Action & Observation.State)

  • 维度: 12维
  • 数据类型: float32
  • 字段名称:
    • left_shoulder_pan.pos
    • left_shoulder_lift.pos
    • left_elbow_flex.pos
    • left_wrist_flex.pos
    • left_wrist_roll.pos
    • left_gripper.pos
    • right_shoulder_pan.pos
    • right_shoulder_lift.pos
    • right_elbow_flex.pos
    • right_wrist_flex.pos
    • right_wrist_roll.pos
    • right_gripper.pos

图像观测(Observation.Images)

  • 三个摄像头视角:
    • wrist_left: 分辨率 480×640,3通道,视频编码 av1,帧率 10 FPS
    • center: 分辨率 480×640,3通道,视频编码 av1,帧率 10 FPS
    • wrist_right: 分辨率 480×640,3通道,视频编码 av1,帧率 10 FPS
  • 是否深度图: 否
  • 含音频: 否

其他字段

  • timestamp: float32,形状 [1]
  • frame_index: int64,形状 [1]
  • episode_index: int64,形状 [1]
  • index: int64,形状 [1]
  • task_index: int64,形状 [1]

引用信息

  • BibTeX: 当前为 [More Information Needed]
二维码
社区交流群
二维码
科研交流群
商业服务