遇见数据集

rq4-pap-four-objects-275-hsv-sam-half-seed42-normalized

收藏
Hugging Face2026-06-08 更新2026-06-08 收录
官方服务:

资源简介:

这是一个机器人学数据集,使用LeRobot框架创建。数据集专注于机器人类型so_follower,包含275个episodes,总计92173帧数据,帧率为30fps。数据以parquet格式存储,总数据文件大小为100MB,视频文件大小为200MB。数据集仅包含训练集(splits为train: 0:275)。特征包括:动作(action),为6维浮点数组,表示机器人的关节位置(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);观察状态(observation.state),同样为6维浮点数组,表示机器人的当前状态;图像观察(observation.images.eagle和observation.images.side),为视频格式,分辨率240x320,3通道,来自两个摄像头视角;以及元数据如时间戳、帧索引、episode索引、索引和任务索引。数据集适用于机器人控制、模仿学习或强化学习任务。

This is a robotics dataset created using the LeRobot framework. It focuses on the so_follower robotic platform, containing 275 episodes with a total of 92,173 frames at a frame rate of 30 fps. The dataset is stored in Parquet format, with a total data file size of 100 MB and video files totaling 200 MB. The dataset only includes the training split, with splits configured as train: 0:275. The features included are as follows: 1. Action: a 6-dimensional floating-point array representing the robot's joint positions, including shoulder translation, shoulder lift, elbow flexion, wrist flexion, wrist rotation, and gripper position; 2. Observation state (observation.state): also a 6-dimensional floating-point array representing the robot's current operational state; 3. Image observations (observation.images.eagle and observation.images.side): stored in video format, with a resolution of 240×320 and 3 color channels, captured from two separate camera viewpoints; 4. Metadata such as timestamps, frame indices, episode indices, sample indices, and task indices. This dataset is suitable for robotics control, imitation learning, and reinforcement learning tasks.

提供机构:
martonbodo
创建时间:
2026-06-08
原始信息汇总

数据集概述

  • 名称:rq4-pap-four-objects-275-hsv-sam-half-seed42-normalized
  • 许可证:Apache-2.0
  • 任务类别:机器人学(Robotics)
  • 标签:LeRobot

数据集来源

数据集规模与结构

属性
机器人类型 so_follower
总片段数(Episodes) 275
总帧数(Frames) 92,173
总任务数(Tasks) 1
数据块大小(Chunk Size) 1,000
数据文件大小 100 MB
视频文件大小 200 MB
帧率(FPS) 30
划分(Splits) 训练集:第0至274片段(全部用于训练)

数据存储格式

  • 数据文件路径data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
  • 视频文件路径videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
  • 元数据文件meta/info.json

特征(Features)

特征名 数据类型 形状(Shape) 说明
action float32 [6] 包含6个关节的动作:shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, gripper.pos
observation.state float32 [6] 机器人状态(6个关节位置),与action名称相同
observation.images.eagle video [240, 320, 3] “鹰眼”视角摄像机视频,分辨率240×320,AV1编码,30 FPS,无音频
observation.images.side video [240, 320, 3] 侧视角摄像机视频,分辨率240×320,AV1编码,30 FPS,无音频
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

视频信息

  • 高度:240 像素
  • 宽度:320 像素
  • 通道数:3(RGB)
  • 编码格式:AV1
  • 像素格式:yuv420p
  • 是否为深度图:否
  • 帧率:30 FPS
  • 是否有音频:否
二维码
社区交流群
二维码
科研交流群
商业服务