autoresearch-000114-cucumber-grab
收藏资源简介:
这是一个使用LeRobot创建的机器人数据集,专为机器人控制任务设计。数据集包含一个总时长为147帧的单个任务(总任务数为1),帧率为20 FPS。数据特征包括:观测状态(7个关节位置:肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转、夹爪位置和倾斜位置)、三个视角的视频观测(顶部、腕部和基部,每个视频分辨率为480x640,3通道,使用h264编码)、动作数据(与观测状态相同的7个关节位置)、以及时间戳、帧索引、任务索引和剧集索引等元数据。数据集采用Apache 2.0许可证,机器人类型为vibeboard_follower_tilt,数据以Parquet文件格式存储(用于结构化数据),视频以MP4格式存储。数据集结构支持训练分割(train: 0:1),总数据文件大小为100 MB,视频文件大小为200 MB。
This is a robot dataset created using LeRobot, specifically designed for robot control tasks. The dataset contains a single task with a total duration of 147 frames (total task count: 1), at a frame rate of 20 FPS. Data features include: observation state (7 joint positions: shoulder translation, shoulder lift, elbow bend, wrist bend, wrist rotation, gripper position, and tilt position), video observations from three perspectives (top, wrist, and base, each with a resolution of 480x640, 3 channels, encoded using h264), action data (the same 7 joint positions as the observation state), and metadata such as timestamps, frame index, task index, and episode index. The dataset is licensed under Apache 2.0, with a robot type of vibeboard_follower_tilt. Data is stored in Parquet file format (for structured data), and videos are stored in MP4 format. The dataset structure supports a training split (train: 0:1), with a total data file size of 100 MB and a video file size of 200 MB.
数据集概述
- 名称: autoresearch-000114-cucumber-grab
- 许可证: Apache-2.0
- 任务类别: 机器人学 (Robotics)
- 标签: LeRobot
- 创建工具: LeRobot (https://github.com/huggingface/lerobot)
数据集结构
- 格式: Parquet、MP4
- 帧率 (FPS): 20
- 总片段数: 1
- 总帧数: 147
- 总任务数: 1
- 机器人类型: vibeboard_follower_tilt
- 数据文件大小: 约 100 MB
- 视频文件大小: 约 200 MB
特征信息
| 特征名称 | 数据类型 | 形状 | 描述 |
|---|---|---|---|
| observation.state | float32 | [7] | 7维状态:肩膀旋转、肩膀抬升、肘弯曲、腕弯曲、腕滚动、夹爪、倾斜位置 |
| observation.images.top | 视频 (h264) | 480x640x3 | 顶部摄像头 RGB 图像 |
| observation.images.wrist | 视频 (h264) | 480x640x3 | 腕部摄像头 RGB 图像 |
| observation.images.base | 视频 (h264) | 480x640x3 | 底部摄像头 RGB 图像 |
| action | float32 | [7] | 7维动作:与状态维度对应 |
| 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 |
划分
- 训练集: 0:1 (全部数据用于训练)



