DL_project_box_fail
收藏资源简介:
该数据集使用LeRobot平台创建,是一个机器人学领域的数据集,适用于机器人学习任务(如模仿学习或强化学习)。数据集包含一个名为so_follower的机器人的交互数据。数据规模总计40个情节(episodes),28895帧,涵盖1个任务。数据以Parquet文件格式存储,总数据文件大小约为100 MB,关联的视频文件大小约为200 MB,视频帧率为30 fps。数据集仅包含训练集划分(train: 0:40)。数据集的结构化特征包括:动作(action),是一个6维浮点向量,表示机器人六个关节(肩部平移、肩部升降、肘部弯曲、腕部弯曲、腕部旋转、夹爪)的位置指令;观察状态(observation.state),是一个6维浮点向量,表示与动作相同的六个关节的当前位置;两个图像观察:observation.images.handeye(手眼摄像头视角)和observation.images.side(侧面摄像头视角),均为RGB视频数据,分辨率360x640,3通道,使用AV1编解码器;此外还包括时间戳(timestamp)、帧索引(frame_index)、情节索引(episode_index)、索引(index)和任务索引(task_index)等元数据字段。数据按块组织,便于高效访问。
This dataset was created using the LeRobot platform and is a dataset in the field of robotics, suitable for robot learning tasks (such as imitation learning or reinforcement learning). The dataset contains interaction data from a robot named so_follower. The data scale totals 40 episodes, 28,895 frames, covering 1 task. The data is stored in Parquet file format, with a total data file size of approximately 100 MB, and associated video files of about 200 MB, with a video frame rate of 30 fps. The dataset only includes a training set split (train: 0:40). The structured features of the dataset include: action, which is a 6-dimensional floating-point vector representing position commands for the robots six joints (shoulder translation, shoulder lift, elbow bend, wrist bend, wrist rotation, gripper); observation.state, which is a 6-dimensional floating-point vector representing the current positions of the same six joints as the action; two image observations: observation.images.handeye (hand-eye camera view) and observation.images.side (side camera view), both RGB video data with a resolution of 360x640, 3 channels, using the AV1 codec; additionally, metadata fields such as timestamp, frame_index, episode_index, index, and task_index are included. The data is organized in chunks for efficient access.
数据集概述:DL_project_box_fail
- 数据集地址:https://huggingface.co/datasets/yywang122/DL_project_box_fail
- 许可协议:Apache-2.0
- 任务类别:机器人学(robotics)
- 标签:LeRobot
数据集规模与结构
- 总片段数(episodes):40
- 总帧数(frames):28,895
- 总任务数:1
- 数据分块:每块包含1000帧(chunks_size: 1000)
- 数据文件大小:约100 MB(parquet格式)
- 视频文件大小:约200 MB(mp4格式)
- 帧率(fps):30
- 数据划分:所有40个片段均用于训练(splits.train: "0:40")
机器人类型
- 机器人名称:so_follower
数据特征(features)
| 特征名称 | 数据类型 | 维度/形状 | 说明 |
|---|---|---|---|
action |
float32 | (6,) | 6维关节动作:shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper(位置) |
observation.state |
float32 | (6,) | 6维关节状态,与动作维度相同 |
observation.images.handeye |
视频(video) | (360, 640, 3) | 手眼相机图像,分辨率360×640,编码av1,30fps,无音频 |
observation.images.side |
视频(video) | (360, 640, 3) | 侧方相机图像,分辨率360×640,编码av1,30fps,无音频 |
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(仅1个任务) |
数据存储路径
- 数据文件:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频文件:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
引用信息
- BibTeX:暂未提供([More Information Needed])




