遇见数据集

libero_object_iiwa_additionalCams_failures

收藏
Hugging Face2026-06-13 更新2026-06-13 收录
官方服务:

资源简介:

该数据集使用LeRobot创建,专注于机器人学任务,具体配置为libero_object_iiwa_additionalCams_failures。数据集包含iiwa机器人的操作数据,总共有43个episodes、6964帧和9个不同任务。数据以parquet格式存储,特征包括观察状态(如电机状态、末端执行器状态、关节状态和夹爪状态)、多视角图像(前视图和侧视图,分辨率为256x256 RGB视频,帧率为20fps)、动作数据(7维浮点数组)以及时间戳、帧索引、episode索引等元数据。数据集主要用于训练和评估机器人控制算法,特别关注失败案例的额外摄像头记录。

This dataset is created using LeRobot, focusing on robotics tasks with the specific configuration named libero_object_iiwa_additionalCams_failures. It contains operation data from the iiwa robot, totaling 43 episodes, 6964 frames, and 9 distinct tasks. The dataset is stored in Parquet format, with features including observation states (e.g., motor states, end-effector states, joint states, and gripper states), multi-view images (front and side views, 256×256 RGB videos at 20 fps), action data (7-dimensional floating-point arrays), as well as metadata such as timestamps, frame indices, and episode indices. This dataset is mainly used for training and evaluating robotic control algorithms, with a particular focus on the additional camera recordings of failure cases.

提供机构:
OliverHausdoerfer
创建时间:
2026-06-13
原始信息汇总

数据集概述

  • 数据集名称: libero_object_iiwa_additionalCams_failures
  • 许可证: Apache-2.0
  • 任务类别: 机器人学
  • 创建工具: LeRobot
  • 机器人类型: iiwa

数据集规模

  • 总片段数: 43
  • 总帧数: 6964
  • 总任务数: 9
  • 数据文件大小: 100 MB
  • 视频文件大小: 200 MB
  • 帧率: 20 FPS
  • 分片大小: 1000
  • 数据拆分: 全部片段(0:43)用于训练

数据特征

观测状态(observation.state)

  • 数据类型: float32
  • 形状: [8]
  • 电机名称: ["x", "y", "z", "axis_angle1", "axis_angle2", "axis_angle3", "gripper", "gripper"]

末端执行器状态(observation.states.ee_state)

  • 数据类型: float32
  • 形状: [6]
  • 电机名称: ["x", "y", "z", "axis_angle1", "axis_angle2", "axis_angle3"]

关节状态(observation.states.joint_state)

  • 数据类型: float32
  • 形状: [7]
  • 电机名称: ["joint_0", "joint_1", "joint_2", "joint_3", "joint_4", "joint_5", "joint_6"]

夹爪状态(observation.states.gripper_state)

  • 数据类型: float32
  • 形状: [2]
  • 电机名称: ["gripper", "gripper"]

图像观测

  • 前视图像(observation.images.frontview_image):
    • 数据类型: 视频
    • 形状: [256, 256, 3]
    • 编码: AV1, 20 FPS, RGB
  • 侧视图像(observation.images.sideview_image):
    • 数据类型: 视频
    • 形状: [256, 256, 3]
    • 编码: AV1, 20 FPS, RGB

动作(action)

  • 数据类型: float32
  • 形状: [7]
  • 电机名称: ["x", "y", "z", "axis_angle1", "axis_angle2", "axis_angle3", "gripper"]

其他特征

  • 时间戳(timestamp): float32,形状 [1]
  • 帧索引(frame_index): int64,形状 [1]
  • 片段索引(episode_index): int64,形状 [1]
  • 全局索引(index): int64,形状 [1]
  • 任务索引(task_index): int64,形状 [1]

数据格式

  • 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
二维码
社区交流群
二维码
科研交流群
商业服务