jones_jul10_tabletop_teleop_auto_left_10hz

该数据集是一个机器人数据集，使用LeRobot工具创建，包含1007个episodes和141869个frames，数据以parquet格式存储，视频以mp4格式存储。数据集的特征包括时间戳、帧索引、观测状态（如关节状态、力传感器数据、图像等）、动作命令（如关节命令、末端执行器目标等）以及成功标志。机器人类型为agibot_g1，帧率为10Hz。This is a robotic dataset developed using the LeRobot toolkit. It comprises 1007 episodes and 141,869 frames. The dataset itself is stored in Parquet format, while the associated video files are saved in MP4 format. The dataset features include timestamps, frame indices, observation states (e.g., joint states, force sensor data, images), action commands (e.g., joint commands, end-effector targets), and success flags. The robotic platform used for this dataset is agibot_g1, with a frame rate of 10 Hz.

原始信息汇总

数据集概述

• 数据集名称: avatarrobotics/jones_jul10_tabletop_teleop_auto_left_10hz
• 许可证: Apache-2.0
• 任务类别: 机器人学 (Robotics)
• 创建工具: LeRobot

数据集规模

• 总片段数 (Episodes): 1007
• 总帧数 (Frames): 141869
• 总任务数 (Tasks): 1
• 帧率 (FPS): 10
• 数据文件大小: 100 MB
• 视频文件大小: 200 MB

数据集分割

• 训练集 (Train): 片段索引 0 至 1006 (共 1007 个片段)

机器人类型

• 机器人型号: agibot_g1

数据特征

数据集包含以下特征字段，涵盖状态、动作、图像、力觉、位姿等多种模态：

• 时间戳与索引:

  • timestamp (float32, shape [1])
  • frame_index (int64, shape [1])
  • episode_index (int64, shape [1])
  • index (int64, shape [1])
  • task_index (int64, shape [1])

• 观测状态:

  • observation.state (float32, shape [8]): 7个关节状态 + 1个真空吸盘状态
  • observation.state_other (float32, shape [7]): 右臂7个关节状态
  • observation.velocity (float32, shape [7]): 7个关节速度
  • observation.ee_pose (float32, shape [7]): 末端执行器位姿 (位置 xyz + 四元数 xyzw)
  • observation.body_state (float32, shape [4]): 机器人本体状态 (腰部升降、腰部俯仰、头部偏航、头部俯仰)
  • observation.force (float32, shape [6]): 六维力/力矩 (fx, fy, fz, tx, ty, tz)

• 动作指令:

  • action (float32, shape [8]): 7个关节指令 + 1个真空吸盘指令
  • action.ee_target (float32, shape [7]): 末端执行器目标位姿
  • action.body_command (float32, shape [4]): 机器人本体指令 (腰部升降、腰部俯仰、头部偏航、头部俯仰)

• 图像观测:

  • observation.images.camera (video, 800x1280x3, 10fps, h264编码)
  • observation.images.wrist (video, 720x1280x3, 10fps, h264编码)

• 任务完成标志:

  • next.success (bool, shape [1]): 是否成功