test_tube_0729
收藏资源简介:
该数据集使用LeRobot框架创建,专注于机器人操作任务。数据集包含由双机械臂系统(具体型号为bi_flexiv_rizon4_rt)采集的115个演示轨迹(episodes),共计705,709帧,所有数据均来自单一任务。数据以30fps帧率记录,包含20维的动作指令(action)和观测状态(observation.state),分别对应左右机械臂的工具中心点(TCP)位置(x, y, z)和姿态(r1-r6),以及左右夹爪的位置(gripper.pos)。此外,观测数据还包括多视角视觉信息:头相机、左右手腕相机以及左右触觉传感器各两个(共7个视频流),所有视频均为彩色(3通道),其中头相机和手腕相机分辨率为480×640,触觉传感器分辨率为400×700,均采用h264编码。数据集还包含时间戳、帧索引、episode索引、全局索引和任务索引等辅助字段。数据以Parquet文件(分块存储)和MP4视频文件形式组织,总数据文件大小约100MB,视频文件大小约200MB。该数据集适用于机器人模仿学习、行为克隆、强化学习等任务,特别适用于双机械臂精细操作场景。许可证为Apache-2.0。
This dataset is created using the LeRobot framework, focusing on robot manipulation tasks. It contains 115 demonstration episodes collected by a dual-arm robotic system (specifically the bi_flexiv_rizon4_rt), totaling 705,709 frames, all from a single task. Data is recorded at 30fps, including 20-dimensional action commands and observation states, corresponding to the tool center point (TCP) positions (x, y, z) and orientations (r1-r6) of the left and right arms, as well as the left and right gripper positions. Additionally, observations include multi-view visual information: a head camera, left and right wrist cameras, and two tactile sensors (left and right), totaling 7 video streams. All videos are color (3 channels), with head and wrist camera resolutions of 480×640, tactile sensor resolutions of 400×700, encoded using h264. The dataset also includes auxiliary fields such as timestamps, frame indices, episode indices, global indices, and task indices. Data is organized as Parquet files (chunked storage) and MP4 video files, with total data file size approximately 100MB and video file size approximately 200MB. This dataset is suitable for robot imitation learning, behavior cloning, reinforcement learning, and particularly for fine manipulation tasks with dual arms. License: Apache-2.0.
数据集概述:Xense/test_tube_0729
基本信息
- 许可证:Apache 2.0
- 任务类型:机器人学(Robotics)
- 创建工具:LeRobot(Hugging Face 的机器人学习框架)
- 机器人类型:bi_flexiv_rizon4_rt(双臂 Flexiv Rizon 4 机器人)
数据集规模
| 指标 | 数值 |
|---|---|
| 总片段数(Episodes) | 130 |
| 总帧数(Frames) | 817,681 |
| 任务总数 | 1 |
| 块大小(Chunk Size) | 1000 |
| 数据文件大小 | 100 MB |
| 视频文件大小 | 200 MB |
| 帧率(FPS) | 30 |
| 训练/测试划分 | Train: 0-130(全部用于训练) |
数据特征
动作与状态(Action & Observation State)
- 数据类型:float32
- 维度:20 维
- 包含内容:
- 左/右机械臂 TCP 位姿(x, y, z, r1-r6)
- 左/右夹爪位置(gripper position)
视觉观测(Images)
| 相机视角 | 分辨率 | 编码 | 帧率 |
|---|---|---|---|
| 头部相机 | 480×640×3 | H.264 | 30 FPS |
| 左手腕相机 | 480×640×3 | H.264 | 30 FPS |
| 右手腕相机 | 480×640×3 | H.264 | 30 FPS |
| 左触觉传感器 0 | 400×700×3 | H.264 | 30 FPS |
| 左触觉传感器 1 | 400×700×3 | H.264 | 30 FPS |
| 右触觉传感器 0 | 400×700×3 | H.264 | 30 FPS |
| 右触觉传感器 1 | 400×700×3 | H.264 | 30 FPS |
附加元数据
- timestamp:时间戳(float32)
- frame_index:帧索引(int64)
- episode_index:片段索引(int64)
- index:索引(int64)
- task_index:任务索引(int64)
数据格式
- 数据文件:Parquet 格式,路径为
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频文件:MP4 格式,路径为
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
引用信息
该数据集基于 LeRobot-Xense 项目创建(GitHub: https://github.com/Vertax42/lerobot-xense),支持 Xense 触觉机器人。引用格式见数据集的 BibTeX 条目。




