遇见数据集

demo_src_three_piece_assembly_task_D1_robot_Panda_gripper_PandaGripper_additionalCams

收藏
Hugging Face2026-06-22 更新2026-06-22 收录
官方服务:

资源简介:

该数据集是一个使用LeRobot创建的机器人学数据集,专注于跨具身机器人任务。它包含500个训练episodes,总计167075帧,帧率为20fps。数据以parquet文件格式存储,视频以mp4格式存储。数据集特征包括:动作数据(7维浮点数,表示机器人在x、y、z方向的位移、旋转位移和夹爪控制);观察数据,包括状态(2维浮点数)、末端执行器位置和姿态(7维浮点数)、夹爪状态(6维浮点数)、关节位置(7维浮点数),以及组合特征如末端执行器夹爪状态(13维浮点数)和关节夹爪状态(13维浮点数)。此外,还提供来自frontview和sideview的图像观察,均为256x256分辨率的RGB视频。数据集结构支持机器人学习和控制任务的研究与应用。

This dataset is a robotics dataset developed using LeRobot, focusing on cross-embodied robotic tasks. It contains 500 training episodes, totaling 167,075 frames with a frame rate of 20 fps. The data is stored in Parquet file format, while the videos are saved in MP4 format. The dataset includes the following features: action data (7-dimensional floating-point numbers, representing the robot's displacements in x, y, z directions, rotational displacements, and gripper control); observation data, which includes state (2-dimensional floating-point numbers), end-effector position and pose (7-dimensional floating-point numbers), gripper state (6-dimensional floating-point numbers), joint positions (7-dimensional floating-point numbers), as well as combined features such as end-effector gripper state (13-dimensional floating-point numbers) and joint gripper state (13-dimensional floating-point numbers). Additionally, image observations from front view and side view are provided, both as RGB videos with a resolution of 256×256. The structure of this dataset supports research and applications for robotic learning and control tasks.

提供机构:
OliverHausdoerfer
创建时间:
2026-06-22
原始信息汇总

数据集概述

  • 数据集名称: demo_src_three_piece_assembly_task_D1_robot_Panda_gripper_PandaGripper_additionalCams
  • 许可证: Apache-2.0
  • 任务类型: 机器人学 (Robotics)
  • 创建工具: LeRobot

数据集规模与结构

  • 总集数: 500
  • 总帧数: 167,075
  • 总任务数: 1
  • 帧率: 20 FPS
  • 数据文件大小: 100 MB
  • 视频文件大小: 200 MB
  • 数据分块大小: 1,000 (chunks_size)
  • 数据拆分: 训练集包含所有500集 (索引 0:500)
  • 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
  • 机器人类型: cross_embodiment

特征与数据格式

特征名称 数据类型 形状 字段描述
action float32 [7] 动作指令,包括 dx, dy, dz, drx, dry, drz, gripper
observation.state float32 [2] 机器人状态,包含 zero0, zero1
observation.eef float32 [7] 末端执行器位姿,包含位置 (x, y, z) 和四元数 (x, y, z, w)
observation.gripper float32 [6] 夹爪关节位置,6个自由度
observation.joint float32 [7] 机器人关节位置,7个自由度
observation.eef_gripper float32 [13] 末端执行器位姿与夹爪位置的拼接 (7+6)
observation.joint_gripper float32 [13] 关节位置与夹爪位置的拼接 (7+6)
observation.images.frontview video [256, 256, 3] 前视摄像头 RGB 视频,AV1 编码,20 FPS
observation.images.sideview video [256, 256, 3] 侧视摄像头 RGB 视频,AV1 编码,20 FPS
timestamp float32 [1] 时间戳
frame_index int64 [1] 帧索引
episode_index int64 [1] 集索引
index int64 [1] 全局索引
task_index int64 [1] 任务索引

可视化工具

可通过 Hugging Face Space 可视化该数据集。

二维码
社区交流群
二维码
科研交流群
商业服务