VINE2_poc
收藏资源简介:
这是一个机器人学数据集,使用Physical AI Tools和LeRobot工具创建,专门针对ffw_sg2_rev1机器人类型。数据集包含4个训练集、529帧数据和12个视频,数据以parquet格式存储。数据集中包括来自头部摄像头和左右手腕摄像头的视频观察(分别分辨率为376x672和240x424),以及机器人状态(如关节位置)和动作(如关节控制)的浮点数据。所有数据以10帧每秒的速率采集,适用于机器人控制和学习任务。
This is a robotics dataset created using Physical AI Tools and LeRobot tools, specifically for the ffw_sg2_rev1 robot type. The dataset includes 4 training sets, 529 frames of data, and 12 videos, stored in parquet format. It contains video observations from head and left/right wrist cameras (with resolutions of 376x672 and 240x424, respectively), as well as floating-point data for robot states (e.g., joint positions) and actions (e.g., joint control). All data is collected at a rate of 10 frames per second, suitable for robot control and learning tasks.
数据集概述
- 数据集名称: VINE2_poc
- 许可证: Apache-2.0
- 任务类型: 机器人学 (robotics)
- 标签: robotis, LeRobot, ffw_sg2_rev1
数据集创建
该数据集使用 Physical AI Tools 和 LeRobot 创建。
数据集结构
- 代码库版本: v2.1
- 机器人类型: ffw_sg2_rev1
- 总片段数 (episodes): 4
- 总帧数 (frames): 529
- 总任务数 (tasks): 1
- 总视频数 (videos): 12
- 总分块数 (chunks): 1
- 分块大小: 1000
- 帧率 (fps): 10
- 数据划分: 训练集 (train) 包含所有 4 个片段 (0:4)
数据文件路径
- 数据路径格式:
data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet - 视频路径格式:
videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4
特征 (Features)
| 特征名称 | 数据类型 | 形状 | 说明 |
|---|---|---|---|
timestamp |
float32 | [1] | 时间戳 |
frame_index |
int64 | [1] | 帧索引 |
episode_index |
int64 | [1] | 片段索引 |
index |
int64 | [1] | 索引 |
task_index |
int64 | [1] | 任务索引 |
observation.images.cam_head |
video | [376, 672, 3] | 头部摄像头图像 (高度376, 宽度672, 3通道, 编解码器: libx264, 像素格式: yuv420p) |
observation.images.cam_wrist_left |
video | [240, 424, 3] | 左手腕摄像头图像 (高度240, 宽度424, 3通道, 编解码器: libx264, 像素格式: yuv420p) |
observation.images.cam_wrist_right |
video | [240, 424, 3] | 右手腕摄像头图像 (高度240, 宽度424, 3通道, 编解码器: libx264, 像素格式: yuv420p) |
observation.state |
float32 | [8] | 机器人状态 (arm_l_joint1至arm_l_joint7, gripper_l_joint1) |
action |
float32 | [8] | 动作 (arm_l_joint1至arm_l_joint7, gripper_l_joint1) |



