Pick_and_Place

该数据集是一个由LeRobot创建的机器人操作数据集，专注于拾取与放置任务，特别是拾取与放置盒子。它针对Trossen机器人平台（具体为pi05型号），包含一个训练分割。数据集中有1个情节、778帧、1个任务和4个视频。数据以parquet格式存储，视频为mp4格式，帧率为30fps。特征包括动作（一个14维浮点数组，控制左右机械臂的各7个关节）、观测状态（一个14维浮点数组，表示左右机械臂的各7个关节状态）、来自四个摄像头的图像观测（包括高分辨率摄像头、低分辨率摄像头、左手腕摄像头和右手腕摄像头，所有摄像头视频均为480x640分辨率、3通道RGB格式），以及时间戳、帧索引、情节索引、索引和任务索引等元数据。该数据集适用于机器人学习研究，如模仿学习或强化学习，用于训练机器人执行物体操作任务。This dataset is a robotic manipulation dataset developed by LeRobot, focusing on pick-and-place tasks, specifically box pick-and-place operations. It is tailored for the Trossen robotics platform, specifically the pi05 model, and includes a single training split. The dataset contains 1 episode, 778 frames, 1 task, and 4 videos. The data is stored in Parquet format, while the videos are saved in MP4 format with a frame rate of 30 fps. The features include: actions, a 14-dimensional floating-point array that controls the 7 individual joints of the left and right robotic arms respectively; observation states, a 14-dimensional floating-point array representing the joint states of the left and right robotic arms; image observations from four cameras, namely the high-resolution camera, low-resolution camera, left wrist camera, and right wrist camera, all of which have a resolution of 480x640 and adopt 3-channel RGB format; as well as metadata such as timestamps, frame index, episode index, sample index, and task index. This dataset is applicable to robotics learning research such as imitation learning or reinforcement learning, for training robots to perform object manipulation tasks.

原始信息汇总

数据集概述

• 数据集名称: Pick_and_Place
• 发布机构: uaisroboticslab
• 许可证: Apache-2.0
• 任务类别: 机器人学 (Robotics)
• 标签: LeRobot, Pick_and_Place, Pick_and_Place_the_BOX, trossen, pi05
• 创建工具: 使用 LeRobot 创建

---

数据集结构

• 机器人类型: trossen_ai_stationary
• 总片段数 (Episodes): 1
• 总帧数 (Frames): 778
• 总任务数 (Tasks): 1
• 总视频数 (Videos): 4
• 总数据块数 (Chunks): 1
• 数据块大小 (Chunk Size): 1000
• 帧率 (FPS): 30
• 数据存储路径: data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet
• 视频存储路径: videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4

数据集划分 (Splits)

• 训练集 (Train): 包含所有片段 (索引 0:1)

---

特征 (Features)

特征名称	数据类型	形状	说明
action	float32	[14]	动作指令，包含左右各7个关节位置 (joint_0 ~ joint_6)
observation.state	float32	[14]	观察到的状态，包含左右各7个关节位置
observation.images.cam_high	video	[480, 640, 3]	高位摄像头视频，分辨率 640x480，30fps，AV1编码
observation.images.cam_low	video	[480, 640, 3]	低位摄像头视频，分辨率 640x480，30fps，AV1编码
observation.images.cam_left_wrist	video	[480, 640, 3]	左手腕摄像头视频，分辨率 640x480，30fps，AV1编码
observation.images.cam_right_wrist	video	[480, 640, 3]	右手腕摄像头视频，分辨率 640x480，30fps，AV1编码
timestamp	float32	[1]	时间戳
frame_index	int64	[1]	帧索引
episode_index	int64	[1]	片段索引
index	int64	[1]	数据索引
task_index	int64	[1]	任务索引

---

其他信息

• Homepage: 暂无
• Paper: 暂无
• BibTeX 引用: 暂无