cylinder-in-hole-70-annotated
收藏资源简介:
该数据集是一个用于机器人学任务的数据集,专注于圆柱体入孔(cylinder-in-hole)操作。它通过LeRobot平台创建,包含70个episodes,总计45853帧,帧率为30fps。数据以parquet格式存储,并包括动作和观察特征:动作由6个浮点数表示,对应机器人关节位置(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);观察包括状态(同样由6个浮点数表示)和两个视角的图像数据(前视和侧视),每个图像分辨率为480x640像素,3通道,以视频格式编码(AV1编解码器)。数据集还包含时间戳、帧索引、episode索引等元数据。机器人类型为so_follower,主要用于训练和评估机器人控制模型。
This dataset is designed for robotics tasks, focusing on the cylinder-in-hole operation. It is created via the LeRobot platform and consists of 70 episodes, totaling 45,853 frames at a frame rate of 30fps. The data is stored in parquet format and includes action and observation features: actions are represented by 6 floating-point numbers corresponding to robot joint positions (such as shoulder translation, shoulder lift, elbow bend, wrist bend, wrist rotation, and gripper position); observations include state (also represented by 6 floating-point numbers) and image data from two perspectives (front and side views), with each image having a resolution of 480x640 pixels, 3 channels, and encoded in video format (AV1 codec). The dataset also contains metadata such as timestamps, frame indices, and episode indices. The robot type is so_follower, and it is primarily used for training and evaluating robot control models.
数据集概述
- 数据集名称: cylinder-in-hole-70-annotated
- 许可证: Apache-2.0
- 任务类别: 机器人(Robotics)
- 标签: LeRobot, so101, teleop, cylinder-in-hole
数据集结构
- 总片段数(Episodes): 70
- 总帧数: 45,853
- 帧率(FPS): 30
- 总任务数: 1
数据特征
动作(Action)
- 数据类型: float32
- 形状: [6]
- 名称: 包含6个关节位置信息:
- shoulder_pan.pos
- shoulder_lift.pos
- elbow_flex.pos
- wrist_flex.pos
- wrist_roll.pos
- gripper.pos
观测状态(Observation State)
- 数据类型: float32
- 形状: [6]
- 名称: 同动作特征,包含6个关节位置信息
观测图像(Observation Images)
- 前视摄像头(front):
- 分辨率: 480×640 像素,3通道
- 视频编码: AV1
- 帧率: 30 FPS
- 侧视摄像头(side):
- 分辨率: 480×640 像素,3通道
- 视频编码: AV1
- 帧率: 30 FPS
其他特征
- timestamp: float32,形状 [1]
- frame_index: int64,形状 [1]
- episode_index: int64,形状 [1]
- index: int64,形状 [1]
- task_index: int64,形状 [1]
数据文件信息
- 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
- 数据文件大小: 约100 MB
- 视频文件大小: 约200 MB
- 分块大小: 1000
- 机器人类型: so_follower
数据划分
- 训练集(train): 0 到 70 片段(所有数据用于训练)



