遇见数据集

cell1_20260519_mohammed_loge-color-sorting_20260519_180829

收藏
Hugging Face2026-05-22 更新2026-05-22 收录
官方服务:

资源简介:

该数据集是一个机器人操作数据集,使用LeRobot平台创建,专为机器人学习任务设计。数据集包含6个完整的情节(episodes),总计2681帧数据,涉及单一任务。机器人类型为starpilot_yam_gripper,数据以30帧每秒的速率采集。数据集包括动作数据(14维浮点向量,控制两个机械臂的x、y、z位置、rx、ry、rz旋转和抓取宽度)、观测状态数据(32维浮点向量,包括机械臂的编码器角度、IMU加速度和角速度、姿态四元数以及抓取器状态)和多视角视觉数据(来自左腕、右腕和基座的RGB和深度摄像头图像,其中左腕和右腕摄像头分辨率为480x640,基座摄像头分辨率为768x1024)。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,适用于机器人控制、模仿学习或强化学习研究。

This is a robotic manipulation dataset developed using the LeRobot platform, which is specifically designed for robotic learning tasks. The dataset contains 6 full episodes, with a total of 2681 frames of data, focusing on a single task. The robotic platform used is the starpilot_yam_gripper, and the data was collected at a rate of 30 frames per second (FPS). The dataset includes three categories of data: action data, which is a 14-dimensional floating-point vector that controls the x, y, z positions, rx, ry, rz rotations and gripper width of the two robotic arms; observation state data, which is a 32-dimensional floating-point vector including the encoder angles of the robotic arms, IMU acceleration and angular velocity, attitude quaternions and gripper status; and multi-view visual data, which consists of RGB and depth camera images from the left wrist, right wrist and base mount, where the left and right wrist cameras have a resolution of 480x640, and the base camera has a resolution of 768x1024. The data is stored in Parquet format, with a total data file size of 100 MB and a video file size of 200 MB, and is suitable for research in robotic control, imitation learning or reinforcement learning.

提供机构:
nodogoro
创建时间:
2026-05-22
原始信息汇总

数据集概述

  • 数据集名称: cell1_20260519_mohammed_loge-color-sorting_20260519_180829
  • 许可证: Apache-2.0
  • 任务类别: 机器人 (Robotics)
  • 标签: LeRobot
  • 创建工具: 使用 LeRobot 创建

数据集规模

指标 数值
总片段数 6
总帧数 2681
总任务数 1
数据文件大小 100 MB
视频文件大小 200 MB
帧率 (FPS) 30
训练集划分 0:6(所有6个片段均为训练集)

机器人类型

  • 机器人型号: starpilot_yam_gripper

数据集结构

数据存储

  • 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
  • 数据分块大小: 1000 帧/块

特征说明

特征名称 数据类型 形状 描述
action float32 (14,) 包含两条机械臂的末端位置 (x, y, z, rx, ry, rz) 及夹爪宽度,共14维
observation.state float32 (32,) 包含两条机械臂的编码器角度、IMU数据、位姿、夹爪状态等状态信息,共32维
observation.images.left_wrist_0_camera_rgb_image video (480, 640, 3) 左腕部相机RGB图像,H.264编码,30fps
observation.images.left_wrist_0_camera_rgb_image_depth video (480, 640, 3) 左腕部相机深度图(实际为RGB编码),H.264编码,30fps
observation.images.right_wrist_0_camera_rgb_image video (480, 640, 3) 右腕部相机RGB图像,H.264编码,30fps
observation.images.right_wrist_0_camera_rgb_image_depth video (480, 640, 3) 右腕部相机深度图(实际为RGB编码),H.264编码,30fps
observation.images.base_0_camera_rgb_image video (480, 640, 3) 底座0号相机RGB图像,H.264编码,30fps
observation.images.base_1_camera_rgb_image video (768, 1024, 3) 底座1号相机RGB图像,分辨率更高,H.264编码,30fps
timestamp float32 (1,) 时间戳
frame_index int64 (1,) 帧序号
episode_index int64 (1,) 片段序号
index int64 (1,) 索引
task_index int64 (1,) 任务序号
二维码
社区交流群
二维码
科研交流群
商业服务