grey_towel_folding_day_two_attempt2_20260604_132903

该数据集是一个机器人学领域的数据集，专门用于机器人操作任务，特别是与折叠灰色毛巾相关的任务。数据集使用LeRobot框架创建，包含21个episodes，总计56156帧数据，帧率为30fps。数据格式包括parquet文件和视频文件，总数据大小约为100MB，视频文件大小约为200MB。数据集的特征包括：动作数据（12维浮点数组，表示左右机械臂的关节位置，如肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部滚动和夹爪位置）、状态观测（同样为12维关节位置）、以及来自三个不同视角的图像观测（左左爪视角、右右爪视角和右俯视视角），这些图像观测以视频格式存储，分辨率为480x640，3通道，使用AV1编解码器。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。机器人类型为“bi_so_follower”，可能表示一个双机械臂系统。数据集主要用于训练和评估机器人学习模型，适用于模仿学习、强化学习等任务。This dataset is in the field of robotics, specifically designed for robot manipulation tasks, particularly related to folding gray towels. It was created using the LeRobot framework and contains 21 episodes, totaling 56,156 frames at a frame rate of 30fps. The data format includes parquet files and video files, with a total data size of approximately 100MB and video file size of about 200MB. The dataset features include: action data (a 12-dimensional floating-point array representing joint positions of the left and right robotic arms, such as shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist roll, and gripper position), state observations (similarly 12-dimensional joint positions), and image observations from three different perspectives (left claw view, right claw view, and right top-down view), stored in video format with a resolution of 480x640, 3 channels, using the AV1 codec. Additionally, the dataset includes metadata such as timestamps, frame indices, episode indices, and task indices. The robot type is bi_so_follower, likely indicating a dual-arm robotic system. The dataset is primarily used for training and evaluating robot learning models, suitable for tasks like imitation learning and reinforcement learning.

原始信息汇总

数据集概述

• 数据集名称: grey_towel_folding_day_two_attempt2_20260604_132903
• 创建工具: 基于 LeRobot 构建
• 许可证: Apache-2.0
• 任务类别: 机器人 (robotics)

数据集结构

• 总片段数 (Episodes): 21
• 总帧数 (Frames): 56,156
• 总任务数 (Tasks): 1
• 帧率 (FPS): 30
• 机器人类型: bi_so_follower (双臂跟随机器人)

数据组成

• 数据文件大小: 约 100 MB (Parquet 格式)
• 视频文件大小: 约 200 MB (MP4 格式)
• 文件路径:
  • 数据: data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
  • 视频: videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
• 数据划分: 全部 21 个片段用于训练 ("train": "0:21")

特征说明

特征名称	数据类型	形状	描述
action	float32	[12]	12维动作信号，包含左右手臂各6个关节位置 (shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper)
observation.state	float32	[12]	12维状态观测，与动作信号维度一致
observation.images.left_left_jaw	video	[480, 640, 3]	左侧左颚摄像头视频流 (AV1 编码，30fps)
observation.images.right_right_jaw	video	[480, 640, 3]	右侧右颚摄像头视频流 (AV1 编码，30fps)
observation.images.right_topdown	video	[480, 640, 3]	顶部俯视摄像头视频流 (AV1 编码，30fps)
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

数据集用途

该数据集记录了机器人折叠灰色毛巾的操作过程，包含21个演示片段，提供了多视角视觉输入（3个摄像头）和双臂关节状态/动作序列，可用于模仿学习或机器人操作任务的训练与评估。