test-left-gazebo
收藏资源简介:
该数据集由LeRobot创建,是一个用于机器人学研究的移动操作机器人数据集。它记录了名为sobit_home的机器人在仿真环境中的操作数据,包含1个完整episode和85帧数据,采用Apache-2.0许可证。数据集包括多模态观测和动作数据:动作数据涵盖19个关节的控制指令,涉及头部(头盘和头倾斜关节)、身体(升降关节)、左臂(肩部倾斜、上臂滚动、上臂弯曲、肘部、腕部倾斜、腕部滚动关节)和左手(多个手指关节)以及移动基座(x、y、θ坐标)。观测数据包括关节状态(与动作相同的19个关节)、左末端执行器的6维位姿(位置x,y,z和姿态roll,pitch,yaw)以及来自两个RGB摄像头的视频流:头部相机(640x480分辨率)和左手相机(1920x1200分辨率),视频采用h264编码,帧率为6fps。此外,数据集还包含时间戳、帧索引、episode索引等元数据。机器人详细信息表明其为移动操作机器人,具有完整的机械结构(头部、身体、双臂、双手、移动基座)和视觉传感器系统。该数据集适用于机器人模仿学习、强化学习、视觉运动控制等任务的研究与开发。
This dataset was developed by LeRobot, serving as a mobile manipulation robotics dataset for robotics research. It records operational data of the sobit_home robot in a simulated environment, containing 1 complete episode and 85 frames of data, and is licensed under Apache-2.0. The dataset includes multimodal observation and action data: the action data covers control commands for 19 joints, involving the head (head plate and head tilt joints), body (lift joint), left arm (shoulder tilt, upper arm roll, upper arm flexion, elbow joint, wrist tilt, wrist roll joints), left hand (multiple finger joints), and the mobile base (x, y, θ coordinates). The observation data includes joint states (the same 19 joints as in the action space), 6-dimensional pose of the left end-effector (position x, y, z and orientation roll, pitch, yaw), as well as video streams from two RGB cameras: the head camera (resolution 640×480) and the left hand camera (resolution 1920×1200). The videos are encoded with h264 at a frame rate of 6 FPS. Additionally, the dataset contains metadata such as timestamps, frame indices, and episode indices. Detailed specifications of the robot indicate that it is a mobile manipulation robot with a complete mechanical structure (head, body, dual arms, dual hands, and mobile base) and a visual sensor system. This dataset is applicable to the research and development of tasks including robot imitation learning, reinforcement learning, visual-motor control, and other related fields.
数据集概述
- 名称: test-left-gazebo
- 许可证: Apache-2.0
- 任务类别: 机器人学 (Robotics)
- 创建工具: LeRobot (https://github.com/huggingface/lerobot)
- 机器人类型: 移动操作器 (Mobile Manipulator)
- 机器人名称: sobit_home (版本 1.0.0)
数据集规模
- 总片段数: 1
- 总帧数: 85
- 总任务数: 1
- 帧率 (FPS): 6
- 数据文件大小: 100 MB
- 视频文件大小: 200 MB
- 分割: 所有数据用于训练 (train: 0:1)
数据结构
数据路径
- 数据文件:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频文件:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
特征 (Features)
| 特征名称 | 数据类型 | 形状 | 说明 |
|---|---|---|---|
| action | float32 | [19] | 19个关节/底座的动作指令 |
| action.is_fresh | bool | [19] | 每个动作分量是否新鲜 |
| action.delta | float32 | [19] | 动作的增量值 |
| action.delta.is_fresh | bool | [19] | 每个动作增量是否新鲜 |
| observation.state | float32 | [19] | 19个关节/底座的观测状态 |
| observation.state.is_fresh | bool | [19] | 每个状态分量是否新鲜 |
| action.base | float32 | [3] | 底座动作 (x, y, theta) |
| action.base.is_fresh | bool | [3] | 每个底座动作是否新鲜 |
| observation.ee_pose.left | float32 | [6] | 左臂末端执行器位姿 (x, y, z, roll, pitch, yaw) |
| observation.ee_pose.left.delta | float32 | [6] | 左臂末端执行器位姿增量 |
| observation.images.head_camera | video | [3, 480, 640] | 头部摄像头视频 (h264, 6fps) |
| observation.images.hand_left_camera | video | [3, 1200, 1920] | 左手摄像头视频 (h264, 6fps) |
| 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 |
机器人形态 (Morphology)
机器人 sobit_home 包含以下可操作部件:
- 头部 (Head): 2个关节 (head_pan_joint, head_tilt_joint)
- 身体 (Body): 1个关节 (body_lift_joint)
- 左臂 (Arm Left): 6个关节 (shoulder_tilt, upper_roll, upper_flex, elbow, wrist_tilt, wrist_roll)
- 右臂 (Arm Right): 6个关节 (shoulder_tilt, upper_roll, upper_flex, elbow, wrist_tilt, wrist_roll)
- 左端执行器 (End Effector Left): 8个手指关节
- 右端执行器 (End Effector Right): 8个手指关节
- 移动底座 (Mobile Base): 支持前进/后退和转向,有8个车轮关节
传感器
- 类型: RGB摄像头
- 摄像头列表:
- head_camera: 640x480 分辨率
- hand_left_camera: 1920x1200 分辨率
- hand_right_camera: 1920x1200 分辨率
用户信息
- 创建者: Keith Valentin
- 邮箱: e25d5303@soka-u.jp
- 地点: 日本,创价大学
引用
- BibTeX: [More Information Needed]




