five

cube-dataset-60-v2_20260704_114346

收藏
Hugging Face2026-07-04 更新2026-07-04 收录
下载链接:
https://huggingface.co/datasets/simonholmes001/cube-dataset-60-v2_20260704_114346
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集是一个机器人控制数据集,由LeRobot创建,专注于机械臂操作任务。它包含60个episodes,总计26,941帧,帧率为30fps。数据特征包括动作(action)和观测状态(observation.state),动作和状态均包含6个浮点数,分别对应机械臂的关节位置(如肩部平移、肩部提升、肘部弯曲、手腕弯曲、手腕旋转和夹爪位置)。观测部分包括侧视和俯视图像,图像分辨率为480x640,3通道彩色视频,编码格式为h264。此外,数据集还包含时间戳、帧索引、episode索引等元数据。机器人类型为so_follower,数据以Parquet文件格式存储,视频文件以MP4格式存储。该数据集适用于强化学习、模仿学习或机器人控制研究,用于训练和评估机械臂在操作任务中的性能。

This dataset is a robot control dataset created using LeRobot, focusing on robotic arm manipulation tasks. It contains 60 episodes, totaling 26,941 frames, with a frame rate of 30fps. The data features include actions (action) and observation states (observation.state), both consisting of 6 floating-point numbers corresponding to the joint positions of the robotic arm (such as shoulder translation, shoulder lift, elbow bend, wrist bend, wrist rotation, and gripper position). The observation part includes side-view and top-view images, with an image resolution of 480x640, 3-channel color video, encoded in h264 format. Additionally, the dataset contains metadata such as timestamps, frame indices, and episode indices. The robot type is so_follower, and the data is stored in Parquet file format, with video files stored in MP4 format. This dataset is suitable for reinforcement learning, imitation learning, or robot control research, used for training and evaluating the performance of robotic arms in manipulation tasks.
提供机构:
simonholmes001
创建时间:
2026-07-04
原始信息汇总

数据集概述

  • 数据集名称: cube-dataset-60-v2_20260704_114346
  • 许可证: Apache-2.0
  • 任务类别: 机器人学 (robotics)
  • 标签: LeRobot
  • 创建工具: 使用 LeRobot 生成

数据集结构

  • 代码库版本: v3.0
  • 帧率 (FPS): 30
  • 机器人类型: so_follower
  • 总片段数 (Episodes): 60
  • 总帧数 (Frames): 26,941
  • 总任务数 (Tasks): 1
  • 数据分块大小: 1000
  • 数据文件大小: 100 MB
  • 视频文件大小: 200 MB
  • 数据划分:
    • 训练集 (train): 0-60 (全部60个片段)

特征 (Features)

特征名称 数据类型 形状 描述
action float32 [6] 机器人动作指令,包含6个关节位置 (shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper)
observation.state float32 [6] 机器人观测状态,与动作相同的6个关节位置
observation.images.side video [480, 640, 3] 侧视角摄像头视频,分辨率为480x640,RGB三通道,H.264编码,30 FPS
observation.images.top video [480, 640, 3] 顶部视角摄像头视频,分辨率为480x640,RGB三通道,H.264编码,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
二维码
社区交流群
二维码
科研交流群
商业服务