vla0-real-world-dataset
收藏资源简介:
该数据集是一个用于机器人学习的真实世界数据集,使用LeRobot平台创建。数据集包含机器人的动作指令、状态观测、多视角图像(包括手腕、上方、右侧、左侧和中间视角的RGB视频,分辨率为480x640,帧率30fps)、中间视角的深度图、电机电流和速度等传感器数据。数据以6自由度机械臂(包括肩部平移、肩部升降、肘部弯曲、手腕弯曲、手腕旋转和夹爪位置)为控制对象,记录了100个任务片段,总计41258帧数据。数据集采用Apache 2.0许可证,适用于机器人控制、视觉感知和强化学习等研究任务。
This real-world robotics learning dataset was developed using the LeRobot platform. It contains robot action commands, state observations, multi-view RGB videos (from wrist, overhead, right-side, left-side, and center viewpoints, with a resolution of 480×640 and a frame rate of 30fps), depth maps from the center viewpoint, and sensor data including motor current and speed. The dataset takes a 6-degree-of-freedom (6-DoF) robotic arm—covering shoulder translation, shoulder lift, elbow flexion, wrist flexion, wrist rotation, and gripper position—as its control target, and records 100 task segments with a total of 41258 frames. Licensed under Apache 2.0, this dataset is applicable to research tasks such as robot control, visual perception, and reinforcement learning.
数据集概述
数据集名称: vla0-real-world-dataset
数据集地址: https://huggingface.co/datasets/mattpidden/vla0-real-world-dataset
许可证: Apache-2.0
任务类别: 机器人学(robotics)
标签: LeRobot
创建工具: 该数据集使用 LeRobot 创建。
数据集结构
- 格式: 数据存储在 Parquet 文件中,视频存储在 MP4 文件中。
- 文件路径:
- 数据:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
- 数据:
- 数据规模:
- 总片段数(Episodes):100
- 总帧数(Frames):41,258
- 总任务数(Tasks):1
- 数据文件大小:100 MB
- 视频文件大小:200 MB
- 帧率(FPS): 30
- 机器人类型: so_follower
- 数据集拆分:
- 训练集(train):片段 0 到 99
特征(Features)
该数据集包含以下特征:
| 特征名称 | 数据类型 | 形状 | 说明 |
|---|---|---|---|
| action | float32 | [6] | 机器人动作指令,包含6个关节的位置(shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper) |
| observation.state | float32 | [6] | 机器人关节状态(与动作维度一致,共6个关节位置) |
| observation.images.wrist | video | [480, 640, 3] | 腕部摄像头视频数据,分辨率480x640,RGB三通道,AV1编码 |
| observation.images.above | video | [480, 640, 3] | 上方摄像头视频数据,分辨率480x640,RGB三通道,AV1编码 |
| observation.images.right | video | [480, 640, 3] | 右侧摄像头视频数据,分辨率480x640,RGB三通道,AV1编码 |
| observation.images.left | video | [480, 640, 3] | 左侧摄像头视频数据,分辨率480x640,RGB三通道,AV1编码 |
| observation.images.middle | video | [480, 640, 3] | 中间摄像头视频数据,分辨率480x640,RGB三通道,AV1编码 |
| observation.motor_currents | int32 | [6] | 6个关节的电机电流值(shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper) |
| observation.motor_velocities | int32 | [6] | 6个关节的电机速度值(shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper) |
| observation.depths.middle | uint16 | [480, 640] | 中间摄像头的深度图,分辨率480x640 |
| 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 |
其他信息
- 主页: 未提供
- 论文: 未提供
- 引用: 未提供 BibTeX 引用信息
- 代码库版本: v3.0



