遇见数据集

lift_red_cube_semi_max_v2

收藏
Hugging Face2026-07-11 更新2026-07-11 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数据集,专注于机械臂操作,特别是通过LeRobot框架创建。数据集包含机械臂的动作和观测数据:动作数据涉及6个关节位置(肩部平移、肩部升降、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置),观测数据包括机械臂状态(同样6个关节位置)和两个视角的视频图像(前置摄像头和整体视角,分辨率为480x640,每秒30帧)。数据以Parquet文件格式存储,总共有10个回合、1904帧,适用于训练机器人控制模型,支持视觉-动作映射任务。数据集结构还包括时间戳、帧索引、回合索引等元数据,适用于机器人学习和模拟环境中的任务执行。

This dataset is intended for robotics tasks, with a focus on robotic arm manipulation, and was specifically developed using the LeRobot framework. It comprises two types of data related to robotic arms: action data and observation data. The action data covers 6 joint positions, namely shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist rotation, and gripper position. The observation data includes robotic arm states (also consisting of the aforementioned 6 joint positions) and video images captured from two perspectives: the front-facing camera and the overall view, with a resolution of 480×640 and a frame rate of 30 frames per second. The data is stored in Parquet file format, with a total of 10 episodes and 1904 frames. It is suitable for training robotic control models and supports visual-action mapping tasks. Additionally, the dataset structure includes metadata such as timestamps, frame indices, and episode indices, which enables its application to task execution in robot learning and simulation environments.

提供机构:
OctpusVLA
创建时间:
2026-07-11
原始信息汇总

数据集概述:lift_red_cube_semi_max_v2

  • 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 创建

数据集结构

  • 代码库版本:v3.0
  • 帧率:30 FPS
  • 机器人类型:so_follower
  • 总片段数:10
  • 总帧数:1904
  • 总任务数:1
  • 数据文件大小:100 MB
  • 视频文件大小:200 MB
  • 数据集划分:训练集(train)包含所有数据(0:10)

特征(Features)

特征名 数据类型 形状 说明
action float32 [6] 动作空间:6个关节位置(shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, gripper.pos)
observation.state float32 [6] 观测状态:与动作空间相同的6个关节位置
observation.images.front video [480, 640, 3] 前置摄像头视频,分辨率480x640,3通道,AV1编码,30 FPS
observation.images.overall video [480, 640, 3] 全局摄像头视频,分辨率480x640,3通道,AV1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

数据存储路径

  • 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
  • 分块大小:1000

可视化

您可以通过以下链接可视化此数据集:Visualize This Dataset

二维码
社区交流群
二维码
科研交流群
商业服务