遇见数据集

autoresearch-vibepi-000238-cucumber-grab

收藏
Hugging Face2026-07-26 更新2026-07-26 收录
官方服务:

资源简介:

该数据集是一个基于LeRobot工具创建的机器人学习数据集,包含机器人执行任务时的多模态数据。具体数据包括:机器人状态观察,由7个浮点数表示的关节位置(如肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部滚动、夹持器位置和倾斜位置);来自三个视角(base、top、wrist)的视频观察,每个视频分辨率为480x640,3通道,帧率20fps,编码为av1;以及相应的动作数据(7个浮点数表示的关节位置)。此外,数据集还包含时间戳、帧索引、episode索引、任务索引等元数据。数据集规模为1个episode、61帧、1个任务,以parquet格式存储,总数据文件大小约100MB,视频文件大小约200MB,机器人类型为vibeboard_follower_tilt,适用于机器人控制、强化学习等任务。

This dataset is a robot learning dataset created using the LeRobot tool, containing multimodal data from robot task execution. The data includes: robot state observations represented by 7 floating-point numbers for joint positions (e.g., shoulder translation, shoulder lift, elbow flexion, wrist flexion, wrist roll, gripper position, and tilt position); video observations from three perspectives (base, top, wrist), each with a resolution of 480x640, 3 channels, a frame rate of 20fps, and encoded as av1; and corresponding action data (7 floating-point numbers for joint positions). Additionally, the dataset includes metadata such as timestamps, frame indices, episode indices, and task indices. The dataset consists of 1 episode, 61 frames, and 1 task, stored in parquet format with a total data file size of approximately 100MB and video file size of approximately 200MB. The robot type is vibeboard_follower_tilt, and it is suitable for tasks like robot control and reinforcement learning.

提供机构:
VibeCuisine
创建时间:
2026-07-26
原始信息汇总

数据集概述

  • 名称: autoresearch-vibepi-000238-cucumber-grab
  • 许可证: Apache-2.0
  • 任务类别: 机器人学 (robotics)
  • 创建工具: 基于 LeRobot 创建

数据集结构

  • 数据格式: Parquet (.parquet) 和 MP4 视频文件
  • 数据路径: data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
  • 视频路径: videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
  • 帧率 (FPS): 20
  • 机器人类型: vibeboard_follower_tilt

特征说明

特征名称 数据类型 形状 描述
observation.state float32 [7] 机器人关节状态:shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, gripper.pos, tilt.pos
observation.images.base 视频 [480, 640, 3] 基座摄像头图像 (RGB, 480x640, AV1编码)
observation.images.top 视频 [480, 640, 3] 顶部摄像头图像 (RGB, 480x640, AV1编码)
observation.images.wrist 视频 [480, 640, 3] 腕部摄像头图像 (RGB, 480x640, AV1编码)
action float32 [7] 动作指令 (与state维度相同)
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

统计信息

  • 总片段数 (total_episodes): 1
  • 总帧数 (total_frames): 61
  • 总任务数 (total_tasks): 1
  • 数据文件大小: 100 MB
  • 视频文件大小: 200 MB
  • 训练/测试划分: 仅包含训练集 (train: 0:1)

可视化

二维码
社区交流群
二维码
科研交流群
商业服务