rollout_smolvla_put_frozen_eval_20260714_20260713_224436
收藏资源简介:
该数据集是一个机器人操作任务数据集,使用LeRobot工具创建。数据集包含10个总集数,总计7678帧,数据以parquet格式存储,视频以mp4格式存储。数据集的特征包括动作(action)和观察状态(observation.state),动作和观察状态都包含6个浮点数值,分别对应机器人关节位置(如肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置)。观察部分还包括顶部图像(observation.images.top)和腕部图像(observation.images.wrist),均为视频格式,分辨率为720x1280,3个通道,帧率为30fps,使用av1编解码器。其他特征包括时间戳(timestamp)、帧索引(frame_index)、集索引(episode_index)、索引(index)和任务索引(task_index)。机器人类型为so_follower,数据集仅包含训练分割(train)。数据文件总大小为100MB,视频文件总大小为200MB。
This is a robotic manipulation task dataset developed using the LeRobot toolkit. The dataset contains 10 total episodes with a cumulative total of 7678 frames. The tabular data is stored in Parquet format, while the video files are stored in MP4 format. The core features of the dataset include action and observation.state. Both action and observation.state consist of 6 floating-point values, which respectively correspond to the robot's joint positions: shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist rotation, and gripper position. The observation modality also includes top-mounted camera image (observation.images.top) and wrist-mounted camera image (observation.images.wrist), both in video format with a resolution of 720×1280, 3 color channels, a frame rate of 30fps, and encoded with the AV1 codec. Additional features include timestamp, frame_index, episode_index, index, and task_index. The robot type is so_follower, and the dataset only includes the training split (train). The total size of the data files is 100 MB, and the total size of the video files is 200 MB.
数据集概述
- 数据集名称:rollout_smolvla_put_frozen_eval_20260714_20260713_224436
- 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
- 代码库版本:v3.0
- 帧率 (FPS):30
- 机器人类型:so_follower
- 总集数 (Episodes):10
- 总帧数 (Frames):7678
- 总任务数 (Tasks):1
- 数据分块大小:1000
- 数据文件大小:100 MB
- 视频文件大小:200 MB
- 数据路径格式:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频路径格式:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4 - 数据拆分:训练集包含全部10个片段 (0:10)
特征字段
| 字段名称 | 数据类型 | 形状 | 说明 |
|---|---|---|---|
action |
float32 | [6] | 机器人动作,包含肩部俯仰、肩部升降、肘部弯曲、腕部弯曲、腕部滚动、夹爪位置 |
observation.state |
float32 | [6] | 机器人状态,数值名称与动作一致 |
observation.images.top |
video | [720, 1280, 3] | 顶部摄像头视频,分辨率720x1280,AV1编码,30fps |
observation.images.wrist |
video | [720, 1280, 3] | 腕部摄像头视频,分辨率720x1280,AV1编码,30fps |
timestamp |
float32 | [1] | 时间戳 |
frame_index |
int64 | [1] | 帧索引 |
episode_index |
int64 | [1] | 片段索引 |
index |
int64 | [1] | 索引 |
task_index |
int64 | [1] | 任务索引 |
视频参数详情
- 顶部摄像头:非深度图,分辨率720x1280,编码格式av1,像素格式yuv420p,帧率30fps,3通道,无音频
- 腕部摄像头:非深度图,分辨率720x1280,编码格式av1,像素格式yuv420p,帧率30fps,3通道,无音频
其他信息
- 主页:未提供
- 论文:未提供
- 引用信息:未提供



