merged-so101-fold-towel-2camera-exact-stats-unified-prompt
收藏资源简介:
该数据集是一个机器人操作数据集,使用LeRobot工具创建。数据集包含216个训练集片段,总计137883帧,帧率为30fps。数据特征包括:机器人的6个关节位置动作(肩部平移、肩部提升、肘部弯曲、手腕弯曲、手腕旋转、夹爪位置)和对应的状态观测;同时提供来自手腕和顶部摄像头的视频观测,分辨率为480x640,RGB三通道。数据以parquet格式存储,视频文件为MP4格式,总大小约300MB(数据文件100MB,视频文件200MB)。数据集适用于机器人学习任务,如动作预测和视觉感知。
This is a robotic manipulation dataset created using the LeRobot toolkit. The dataset contains 216 training episodes, with a total of 137,883 frames at a frame rate of 30 fps. The data features include the robot's 6-joint position actions (shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist rotation, and gripper position) and their corresponding state observations. Additionally, video observations from the wrist and top-mounted cameras are provided with a resolution of 480 × 640 and three RGB channels. The data is stored in Parquet format, while the video files are in MP4 format, with an overall size of approximately 300 MB (100 MB for the data files and 200 MB for the video files). This dataset is suitable for robotic learning tasks such as action prediction and visual perception.
数据集概述
- 数据集名称:merged-so101-fold-towel-2camera-exact-stats-unified-prompt
- 许可证:Apache-2.0
- 任务类别:机器人学(Robotics)
- 标签:LeRobot
数据集结构
数据集采用 LeRobot 格式,包含以下特征:
| 特征名称 | 数据类型 | 形状 | 描述 |
|---|---|---|---|
action |
float32 | [6] | 机器人动作指令,包括肩部、肘部、手腕及夹爪位置 |
observation.state |
float32 | [6] | 机器人状态观测,与动作维度一致 |
observation.images.wrist |
视频(AV1编码) | [480, 640, 3] | 腕部摄像头图像,30 FPS |
observation.images.top |
视频(AV1编码) | [480, 640, 3] | 顶部摄像头图像,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 |
数据集规模
- 总片段数:216
- 总帧数:137,883
- 任务数量:1
- 数据文件大小:100 MB
- 视频文件大小:200 MB
- 帧率:30 FPS
机器人类型
- 机器人类型:so_follower
数据划分
- 训练集:0 至 215 片段(共 216 个片段,全部用于训练)
其他信息
- 代码库版本:v3.0
- 数据路径:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频路径:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4 - 主页与论文:未提供
- 引用:未提供



