rollout_act_pick_and_place_20260728_214900
收藏资源简介:
该数据集使用LeRobot工具创建,专注于机器人执行拾取放置任务时的动作和观测数据。数据结构包含:动作数据(6个关节位置)、状态观测(6个关节位置)、两个摄像头(基座摄像头和腕部摄像头)的RGB视频(分辨率为480x640,帧率为30fps),以及时间戳、帧索引、回合索引等元数据。数据以Parquet格式存储,包括10个回合、3751帧,总数据量约为300MB(其中数据部分100MB,视频部分200MB)。机器人类型为so_follower,适用于机器人控制与视觉强化学习相关任务。
This dataset is created using the LeRobot tool, focusing on action and observation data when robots perform pick-and-place tasks. Its data structure includes: action data (6 joint positions), state observations (6 joint positions), RGB videos from two cameras (base camera and wrist camera) with a resolution of 480x640 and a frame rate of 30fps, as well as metadata such as timestamps, frame indices, and episode indices. The data is stored in Parquet format, consisting of 10 episodes and 3751 frames, with a total data volume of approximately 300MB (100MB for the non-video data portion and 200MB for the video portion). The robot type is so_follower, which is applicable to tasks related to robot control and visual reinforcement learning.
数据集概述
- 数据集名称:rollout_act_pick_and_place_20260728_214900
- 任务类别:机器人学(Robotics)
- 许可证:Apache-2.0
- 创建工具:LeRobot(https://github.com/huggingface/lerobot)
数据集结构
- 总片段数(Episodes):10
- 总帧数(Frames):3751
- 总任务数(Tasks):1
- 帧率(FPS):30
- 数据文件格式:Parquet(位置:data//.parquet)
- 视频文件格式:MP4(位置:videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4)
- 数据文件总大小:100 MB
- 视频文件总大小:200 MB
- 机器人类型:so_follower
数据特征(Features)
| 特征名称 | 数据类型 | 形状 | 说明 |
|---|---|---|---|
| action | float32 | [6] | 6维动作,包含肩关节、肘关节、腕关节及夹爪位置 |
| observation.state | float32 | [6] | 6维状态(与动作维度一致,为关节位置) |
| observation.images.base_camera | video | [480, 640, 3] | 基础摄像头视频,分辨率480x640,AV1编码,30 FPS |
| observation.images.wrist_camera | video | [480, 640, 3] | 腕部摄像头视频,分辨率480x640,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 |
数据集划分
- 训练集(train):片段索引 0 至 9(共10个片段)
备注
- 该数据集专为机器人抓取与放置(Pick and Place)任务设计,包含从基础摄像头和腕部摄像头捕获的视频观测数据。
- 数据集的代码库版本为 v3.0。
- 更多信息(如主页、论文、引用格式)标记为“需要更多信息”。



