eval_smolvla-so101-4tasks-aug_pick-place_20
收藏Hugging Face2026-06-10 更新2026-06-10 收录
下载链接:
https://huggingface.co/datasets/hjkso1406/eval_smolvla-so101-4tasks-aug_pick-place_20
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是使用LeRobot创建的机器人数据集,专为机器人任务设计。它包含7个episodes,总计10025帧,涉及1个任务。数据以parquet文件格式存储,视频以mp4格式存储。特征包括动作(action)、观察状态(observation.state)、来自两个摄像头的图像(observation.images.camera1和camera2),以及时间戳、帧索引、episode索引等元数据。动作和状态特征均为6维浮点数数组,对应机器人关节位置(如肩部平移、肩部升降、肘部弯曲、手腕弯曲、手腕旋转、夹持器位置)。图像特征为视频格式,分辨率为480x640,3通道,帧率为30fps。数据集适用于机器人控制、视觉导航等研究。
This is a robotics dataset developed using LeRobot, tailored specifically for robotic task research. It consists of 7 episodes totaling 10025 frames, covering one single task. The dataset's data is stored in Parquet file format, while the corresponding videos are stored in MP4 format. The included features are action, observation.state, images from two cameras (observation.images.camera1 and observation.images.camera2), as well as metadata such as timestamps, frame indices, and episode indices. Both the action and state features are 6-dimensional floating-point arrays, which correspond to robot joint positions including shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist rotation, and gripper position. The image features are in video format, with a resolution of 480×640, 3 color channels, and a frame rate of 30 fps. This dataset is suitable for research in fields such as robotic control and visual navigation.
提供机构:
hjkso1406创建时间:
2026-06-10
原始信息汇总
数据集概述:hjkso1406/eval_smolvla-so101-4tasks-aug_pick-place_20
该数据集是一个用于机器人技术领域的评估数据集,基于 LeRobot 框架创建。
基本信息
- 许可证: Apache-2.0
- 任务类别: 机器人 (Robotics)
- 标签: LeRobot
- 机器人类型: so_follower
- 数据集版本: v3.0
数据规模
- 总 Episode 数: 7
- 总帧数: 10,025
- 总任务数: 1
- 数据文件大小: 100 MB
- 视频文件大小: 200 MB
- 帧率 (FPS): 30
- 分割: 训练集 (train) 包含所有 7 个 episode (索引 0 到 6)
数据结构
数据集包含以下特征,并以 Parquet 文件和 MP4 视频文件的形式存储。
特征列表
| 特征名称 | 数据类型 | 形状 | 说明 |
|---|---|---|---|
action |
float32 | [6] | 机器人动作指令,包含 6 个关节位置 (shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper) |
observation.state |
float32 | [6] | 机器人状态观测值,与动作维度一致 (6 个关节位置) |
observation.images.camera1 |
video | [480, 640, 3] | 来自 camera1 的视频流,分辨率 480x640,RGB 3 通道,编码格式 AV1,帧率 30 FPS |
observation.images.camera2 |
video | [480, 640, 3] | 来自 camera2 的视频流,分辨率 480x640,RGB 3 通道,编码格式 AV1,帧率 30 FPS |
timestamp |
float32 | [1] | 时间戳 |
frame_index |
int64 | [1] | 帧索引 |
episode_index |
int64 | [1] | Episode 索引 |
index |
int64 | [1] | 全局索引 |
task_index |
int64 | [1] | 任务索引 |
存储路径
- 数据路径:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频路径:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4



