so101_cube_to_bowl_clean_v1_95eps_20260707_filtered
收藏资源简介:
该数据集使用LeRobot创建,专注于机器人学领域,具体针对so101_follower机器人类型。数据集包含95个episodes,总计23923帧,以30fps的帧率录制。数据以parquet文件格式存储,总数据大小为100MB,视频文件大小为500MB。数据集的特征包括动作(如肩部、肘部、腕部和夹爪的位置)、观测状态(包括固定相机和手眼相机的图像,分辨率为480x640,3通道),以及时间戳、帧索引、episode索引等元数据。所有数据均用于单一任务,并分为训练集。
This dataset is created using LeRobot and focuses on the field of robotics, specifically for the so101_follower robot type. It contains 95 episodes, totaling 23923 frames, recorded at 30fps. The data is stored in parquet format, with a total data size of 100MB and video file size of 500MB. Features of the dataset include actions (such as positions of the shoulder, elbow, wrist, and gripper), observation states (including images from fixed and hand-eye cameras, with a resolution of 480x640 and 3 channels), as well as metadata like timestamps, frame indices, and episode indices. All data is used for a single task and is divided into a training set.
数据集概述
- 数据集名称:
bilileri/so101_cube_to_bowl_clean_v1_95eps_20260707_filtered - 许可证: Apache-2.0
- 任务类别: 机器人学(robotics)
- 标签: LeRobot
- 创建工具: 由 LeRobot 创建
数据集结构
- 机器人类型:
so101_follower - 总片段数(episodes): 95
- 总帧数: 23,923
- 总任务数: 1
- 块大小: 1,000
- 数据文件大小: 100 MB
- 视频文件大小: 500 MB
- 帧率(FPS): 30
- 数据划分: 训练集(train)包含第 0 至 94 条片段(共 95 条)
特征描述
| 特征名称 | 数据类型 | 形状 | 说明 |
|---|---|---|---|
action |
float32 | (6,) | 6 维动作指令,包含肩部、肘部、腕部及夹爪位置 |
observation.state |
float32 | (6,) | 6 维机器人状态,与动作空间一致 |
observation.images.fixed |
video | (480, 640, 3) | 固定视角 RGB 视频,分辨率 480×640,AV1 编码,30 FPS |
observation.images.handeye |
video | (480, 640, 3) | 手眼视角 RGB 视频,分辨率 480×640,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(此处为单一任务) |
数据文件与视频文件路径
- 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



