遇见数据集

so101_hamburger_to_paper_bag_merged_v2

收藏
Hugging Face2026-06-13 更新2026-06-13 收录
官方服务:

资源简介:

该数据集是一个用于机器人学的数据集,专注于双臂机器人(类型为bi_so_follower)执行将汉堡放入纸袋等任务。数据集包含30个回合,总计22329帧,以25帧每秒的速率采集。特征包括动作数据(12维浮点数组,代表左右臂各关节位置和夹爪位置)、观察状态(12维浮点数组,反映机器人状态)、多视角视频观察(如左腕、顶部和右腕视角,每个视频分辨率为480x640,3通道,使用AV1编解码器),以及时间戳、帧索引、回合索引、索引和任务索引等元数据。数据以Parquet格式存储,总数据文件大小为100MB,视频文件大小为200MB,分割为训练集。该数据集旨在支持机器人控制、模仿学习或强化学习任务,通过提供丰富的传感器数据和动作记录来促进算法开发。

This is a robotics dataset focused on tasks such as placing hamburgers into paper bags, executed by a dual-arm robot with the model type bi_so_follower. The dataset contains 30 episodes, with a total of 22329 frames collected at 25 frames per second. Its features include action data (a 12-dimensional floating-point array representing the joint positions and gripper positions of the left and right arms), observation states (a 12-dimensional floating-point array reflecting the robot's status), multi-view video observations (e.g., left wrist, top-down, and right wrist viewpoints; each video has a resolution of 480×640 with 3 color channels, using the AV1 codec), as well as metadata such as timestamps, frame indices, episode indices, general indices, and task indices. The data is stored in Parquet format, with a total data file size of 100 MB and a total video file size of 200 MB, and is split into a training set. This dataset aims to support robotics control, imitation learning, or reinforcement learning tasks, and facilitate algorithm development by providing rich sensor data and action recordings.

提供机构:
mega-joanne
创建时间:
2026-06-13
原始信息汇总

数据集概述

  • 数据集名称:so101_hamburger_to_paper_bag_merged_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

数据集规模

指标 数值
总片段数(总episodes) 30
总帧数 22,329
总任务数 1
数据文件大小 100 MB
视频文件大小 200 MB
帧率(FPS) 25

数据集结构

  • 数据格式:Parquet 文件(data/*/*.parquet
  • 视频格式:MP4 文件(AV1 编码,分辨率 480×640,3 通道 RGB)
  • 分割:训练集包含 30 个片段(train: 0:30

特征字段

特征名称 数据类型 形状 说明
action float32 [12] 左/右双臂共12个关节位置(肩部、肘部、腕部、夹爪)
observation.state float32 [12] 观测状态,与action维度相同
observation.images.left_left_wrist video [480, 640, 3] 左侧左腕相机视频
observation.images.left_top video [480, 640, 3] 左侧顶部相机视频
observation.images.right_right_wrist video [480, 640, 3] 右侧右腕相机视频
timestamp float32 [1] 时间戳
frame_index int64 [1] 帧索引
episode_index int64 [1] 片段索引
index int64 [1] 总体索引
task_index int64 [1] 任务索引(仅1个任务)

机器人类型

  • 机器人型号bi_so_follower(双臂跟随机器人)
二维码
社区交流群
二维码
科研交流群
商业服务