cell1_20260523_mohammed_groceries_packing_20260523_213200
收藏Hugging Face2026-05-24 更新2026-05-24 收录
下载链接:
https://huggingface.co/datasets/nodogoro/cell1_20260523_mohammed_groceries_packing_20260523_213200
下载链接
链接失效反馈官方服务:
资源简介:
该数据集由LeRobot创建,是一个机器人操作数据集,专门用于双机械臂(starpilot_yam_gripper)在杂品打包任务中的演示。数据集包含5个完整的情节(episodes),总计3020帧数据,帧率为30fps。数据以parquet格式存储,并分为训练集。数据集特征包括:动作数据(14维浮点数组,控制两个机械臂的位姿和夹爪宽度)、状态观测数据(32维浮点数组,包括两个机械臂的编码器角度、IMU加速度和角速度、位姿(位置和四元数)、夹爪开合状态和距离)、以及多视角图像观测,包括左腕和右腕的RGB图像和深度图像(分辨率480x640),以及两个基座相机的RGB图像(分辨率分别为480x640和768x1024)。所有图像以视频格式存储,使用h264编解码器。数据集适用于机器人学习、模仿学习或强化学习任务。
This dataset is created by LeRobot and is a robotic manipulation dataset specifically designed for demonstrations of dual robotic arms (starpilot_yam_gripper) in grocery packing tasks. It contains 5 complete episodes, totaling 3020 frames of data at a frame rate of 30fps. The data is stored in parquet format and divided into a training set. Dataset features include: action data (14-dimensional float arrays controlling the pose and gripper width of both arms), state observation data (32-dimensional float arrays including encoder angles, IMU acceleration and angular velocity, pose (position and quaternion), gripper open/close status and distance for both arms), and multi-view image observations, including RGB and depth images from left and right wrists (resolution 480x640), as well as RGB images from two base cameras (resolutions 480x640 and 768x1024). All images are stored in video format using the h264 codec. The dataset is suitable for robotic learning, imitation learning, or reinforcement learning tasks.
提供机构:
nodogoro
创建时间:
2026-05-24
原始信息汇总
数据集概要
- 名称:
nodogoro/cell1_20260523_mohammed_groceries_packing_20260523_213200 - 许可证: Apache-2.0
- 任务类别: 机器人学 (robotics)
- 标签: LeRobot
- 创建方式: 使用 LeRobot 框架创建
数据集结构
- 代码版本: v3.0
- 机器人类型: starpilot_yam_gripper
- 总片段数 (episodes): 5
- 总帧数 (frames): 3020
- 总任务数 (tasks): 1
- 块大小 (chunks_size): 1000
- 数据文件总大小: 100 MB (Parquet格式)
- 视频文件总大小: 200 MB (MP4格式)
- 帧率 (fps): 30
- 数据划分: 全部5个片段用于训练 (train: 0:5)
数据特征
数据包含动作、状态观测、多视角图像、时间戳、索引等信息,具体特征如下:
| 特征名称 | 数据类型 | 形状 | 说明 |
|---|---|---|---|
action |
float32 | [14] | 双臂14维动作:包括 arm1_x, arm1_y, arm1_z, arm1_rx, arm1_ry, arm1_rz, arm1_grip_width, <br>arm2_x, arm2_y, arm2_z, arm2_rx, arm2_ry, arm2_rz, arm2_grip_width |
observation.state |
float32 | [32] | 双臂状态:每臂含编码器角度、IMU加速度/角速度、位姿 (xyz+四元数)、夹爪开合状态及距离 (共16维/臂) |
observation.images.left_wrist_0_camera_rgb_image |
video (h264) | [480, 640, 3] | 左腕RGB相机图像,30 fps |
observation.images.left_wrist_0_camera_rgb_image_depth |
video (h264) | [480, 640, 3] | 左腕深度图(RGB编码),30 fps |
observation.images.right_wrist_0_camera_rgb_image |
video (h264) | [480, 640, 3] | 右腕RGB相机图像,30 fps |
observation.images.right_wrist_0_camera_rgb_image_depth |
video (h264) | [480, 640, 3] | 右腕深度图(RGB编码),30 fps |
observation.images.base_0_camera_rgb_image |
video (h264) | [480, 640, 3] | 基础位姿RGB相机图像(0号),30 fps |
observation.images.base_1_camera_rgb_image |
video (h264) | [768, 1024, 3] | 基础位姿RGB相机图像(1号),30 fps,分辨率更高 |
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 |
数据存储格式
- 数据文件路径:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频文件路径:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
备注
- 所有视频使用H.264编码,像素格式为yuv420p,无音频。
- 数据集首页、论文及引用信息当前标记为“需要更多信息”。



