遇见数据集

dataset-2_20260601_142720

收藏
Hugging Face2026-06-01 更新2026-06-01 收录
官方服务:

资源简介:

这是一个使用LeRobot创建的机器人数据集,专门用于机器人技术任务。数据集包含3个episodes,总计1515帧数据,帧率为30fps。数据特征包括动作(action)和观测状态(observation.state),涉及6个关节位置:肩部平移(shoulder_pan.pos)、肩部升降(shoulder_lift.pos)、肘部弯曲(elbow_flex.pos)、腕部弯曲(wrist_flex.pos)、腕部滚动(wrist_roll.pos)和夹爪位置(gripper.pos)。观测部分还包括来自顶部摄像头(observation.images.top)和腕部摄像头(observation.images.wrist)的视频数据,视频分辨率为480x640,编码为h264,无音频。此外,数据集包含时间戳、帧索引、episode索引、索引和任务索引等元数据。机器人类型为so_follower,数据以parquet文件格式存储,视频以mp4格式存储。数据集适用于训练和评估机器人控制模型,支持强化学习或模仿学习任务。

This is a robotics dataset developed using LeRobot, specifically tailored for robotics technical tasks. The dataset consists of 3 episodes, with a total of 1515 frames at a frame rate of 30fps. The data features include actions (action) and observation states (observation.state), involving 6 joint positions: shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, and gripper.pos. The observation modalities also include video data from the top camera (observation.images.top) and wrist camera (observation.images.wrist), with a resolution of 480x640, encoded in H.264, and no audio track. Additionally, the dataset contains metadata such as timestamps, frame indices, episode indices, indices, and task indices. The robot type is so_follower, with numerical data stored in Parquet file format and videos stored in MP4 format. This dataset is suitable for training and evaluating robotic control models, supporting reinforcement learning or imitation learning tasks.

提供机构:
aenescu
创建时间:
2026-06-01
原始信息汇总

数据集概述

  • 任务类型: 机器人学 (robotics)
  • 许可证: Apache-2.0
  • 创建工具: 使用 LeRobot 创建
  • 可视化: 可通过 可视化链接 查看

数据集结构

  • 版本: v3.0
  • 帧率 (FPS): 30
  • 总片段数: 3
  • 总帧数: 1515
  • 总任务数: 1
  • 机器人类型: so_follower
  • 数据文件大小: 约 100 MB
  • 视频文件大小: 约 200 MB
  • 训练/测试划分: 仅包含训练集 (splits: train "0:3")
  • 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

特征

特征名 类型 形状 说明
action float32 [6] 6维动作,包含肩部回转、肩部升降、肘部弯曲、腕部弯曲、腕部回转和夹爪位置
observation.state float32 [6] 6维观测状态,与动作维度相同
observation.images.top video [480, 640, 3] 顶部摄像头视频,分辨率 480x640,H.264 编码,30 FPS
observation.images.wrist video [480, 640, 3] 腕部摄像头视频,分辨率 480x640,H.264 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
二维码
社区交流群
二维码
科研交流群
商业服务