rollout_eval_stacking_8of10
收藏资源简介:
该数据集是使用LeRobot创建的,专用于机器人学任务,具体涉及一个名为so_follower的机器人类型。数据集包含10个总剧集和5702个总帧数,数据以parquet文件格式存储,视频文件为MP4格式,帧率为30fps。数据集的结构包括动作特征(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)、观察状态(与动作相同的关节位置)、前视和腕部图像(分辨率为480x640的RGB视频),以及时间戳、帧索引、剧集索引等元数据。数据集可能用于机器人动作评估和堆叠任务,标签包括eval和act,表明其适用于评估和行为执行场景。
This dataset was created using LeRobot and is specifically designed for robotics tasks, involving a robot type named so_follower. It contains a total of 10 episodes and 5702 frames, with data stored in parquet file format and video files in MP4 format at a frame rate of 30fps. The dataset structure includes action features (such as shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist rotation, and gripper position), observation states (joint positions identical to actions), front-view and wrist images (RGB videos with a resolution of 480x640), as well as metadata like timestamps, frame indices, and episode indices. The dataset may be used for robot action evaluation and stacking tasks, with labels including eval and act, indicating its suitability for evaluation and behavior execution scenarios.
数据集概述
数据集结构
- 格式: Parquet 文件 (
data/*/*.parquet) - 帧率 (fps): 30
- 总片段数 (episodes): 10
- 总帧数 (frames): 5702
- 总任务数 (tasks): 1
- 数据文件大小: 100 MB
- 视频文件大小: 200 MB
- 机器人类型:
so_follower - 数据划分: 训练集 (train): 片段 0 至 9
特征 (Features)
| 特征名 | 数据类型 | 形状 | 描述 |
|---|---|---|---|
action |
float32 | (6,) | 机器人动作,包含6个关节位置(肩部旋转、肩部升降、肘部弯曲、腕部弯曲、腕部旋转、夹爪位置) |
observation.state |
float32 | (6,) | 机器人状态,与动作维度一致 |
observation.images.front |
video | (480, 640, 3) | 前置摄像头视频,高度480、宽度640、3通道,AV1编码,30 fps |
observation.images.wrist |
video | (480, 640, 3) | 腕部摄像头视频,高度480、宽度640、3通道,AV1编码,30 fps |
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 |
引用
- BibTeX: 尚未提供



