ur5_tube_full_tcp240_converted
收藏资源简介:
该数据集名为ur5_tube_full_tcp240_converted,是一个专门用于机器人操作任务的数据集,针对UR5机械臂。它使用LeRobot工具创建,并采用OpenPI使用的episode-split LeRobot v2.1布局进行存储。核心任务是将一个装有红色内容的实验室试管放入较小的实验室试管架中。数据集包含丰富的特征,包括来自前部、腕部和基座相机的图像和深度视频(分辨率均为480x640x3),以及机器人的状态(7维浮点数)、动作(7维浮点数)、末端执行器姿态(7维浮点数)、力和扭矩观测(6维浮点数)等。数据以parquet文件格式存储每帧的观测和动作,同时提供按episode分割的视频流。元数据显示,数据集共有112个episodes、60,804帧、1个任务、672个视频,帧率为30fps,全部用于训练。
The dataset named ur5_tube_full_tcp240_converted is designed for robot manipulation tasks, specifically for the UR5 robotic arm. It is created using the LeRobot tool and stored in the episode-split LeRobot v2.1 layout used by OpenPI. The core task involves placing a laboratory test tube containing red content into a smaller laboratory test tube rack. The dataset includes rich features such as image and depth videos from front, wrist, and base cameras (all with a resolution of 480x640x3), as well as robot state (7-dimensional floating-point numbers), actions (7-dimensional floating-point numbers), end-effector pose (7-dimensional floating-point numbers), and force and torque observations (6-dimensional floating-point numbers). Data is stored in Parquet format for per-frame observations and actions, along with video streams segmented by episodes. Metadata indicates that the dataset contains 112 episodes, 60,804 frames, 1 task, 672 videos, a frame rate of 30fps, and is entirely used for training.
数据集概述:ur5_tube_full_tcp240_converted
该数据集用于机器人操作任务,基于LeRobot框架创建,采用OpenPI使用的episode-split LeRobot v2.1布局。
任务
- 包含一个任务:将装有红色液体的实验室试管放入小型试管架中。
数据集结构
数据集使用parquet文件存储每帧的机器人观测/动作,并使用episode-split视频存储摄像头流。目录结构如下:
meta/:包含元数据文件(info.json,tasks.jsonl,episodes.jsonl,stats.json,episodes_stats.jsonl)。data/:存储parquet格式的帧数据(例如data/chunk-000/episode_000000.parquet)。videos/:存储每个视频键对应的episode视频(例如videos/chunk-000/<video_key>/episode_000000.mp4)。
特征(Features)
| 特征名称 | 类型 | 形状 | 说明 |
|---|---|---|---|
observation.images.front |
video | [480, 640, 3] | 前方摄像头图像 |
observation.images.wrist |
video | [480, 640, 3] | 腕部摄像头图像 |
observation.images.base |
video | [480, 640, 3] | 基座摄像头图像 |
observation.images.front_depth |
video | [480, 640, 3] | 前方深度图像 |
observation.images.wrist_depth |
video | [480, 640, 3] | 腕部深度图像 |
observation.images.base_depth |
video | [480, 640, 3] | 基座深度图像 |
observation.state |
float32 | [7] | 机器人状态 |
action |
float32 | [7] | 动作指令 |
observation.ee_pose |
float32 | [7] | 末端执行器位姿 |
observation.force_torque |
float32 | [6] | 力/扭矩数据 |
teleop.target_pose |
float32 | [7] | 遥操作目标位姿 |
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 |
元数据
- 代码库版本:v2.1
- 机器人类型:ur5
- 总片段数:112
- 总帧数:60804
- 任务数量:1
- 总视频数:672
- 分块数量:1
- 每块大小:1000
- 帧率(fps):30
- 数据划分:全部112个片段用于训练(train: 0:112)
- 数据路径格式:
data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet - 视频路径格式:
videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4



