pickplace_134cam_newdemos_20260526_161032_remapped
收藏资源简介:
该数据集是使用LeRobot创建的机器人学数据集,专门用于拾放(pickplace)任务。数据集包含100个训练集episodes,总计53178帧,数据以parquet格式存储。特征包括:动作数据(6个浮点值,对应机器人关节位置如肩部、肘部、腕部和夹爪)、观测数据(来自三个相机视角的图像,其中camera1和camera2分辨率为1280x720,camera3为640x480,均为30fps视频,编码为av1,无音频)以及机器人状态(与动作相同的6个浮点值)。此外,还包括episode索引、帧索引、任务索引和时间戳等元数据。数据集总数据文件大小为100MB,视频文件大小为200MB,机器人类型为so_follower,许可证为Apache-2.0。
This dataset is a robotics dataset created using LeRobot, specifically for pick-and-place tasks. It contains 100 training episodes, totaling 53,178 frames, stored in parquet format. Features include: action data (6 floating-point values corresponding to robot joint positions such as shoulder, elbow, wrist, and gripper), observation data (images from three camera views, where camera1 and camera2 have a resolution of 1280x720, camera3 is 640x480, all at 30fps video, encoded as av1 without audio), and robot state (6 floating-point values identical to the actions). Additionally, metadata such as episode index, frame index, task index, and timestamps are included. The total data file size is 100MB, video file size is 200MB, robot type is so_follower, and the license is Apache-2.0.
数据集概述
- 数据集名称:saipuneethgottam/pickplace_134cam_newdemos_20260526_161032_remapped
- 许可证:Apache-2.0
- 任务类别:机器人学(Robotics)
- 标签:LeRobot
数据集来源与工具
- 该数据集使用 LeRobot 创建。
数据集结构
- 数据格式:数据以 Parquet 文件存储,路径为
data/*/*.parquet;视频以 MP4 文件存储,路径为videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4。 - 数据规模:
- 总数据文件大小:约 100 MB
- 总视频文件大小:约 200 MB
- 总帧数:53178 帧
- 总片段数(chunks):1000
- 分割信息:训练集包含索引 0 至 99(共 100 个片段)。
- 帧率:30 FPS
- 机器人类型:so_follower
特征说明
数据集的每个样本包含以下特征:
| 特征名 | 数据类型 | 形状 | 描述 |
|---|---|---|---|
action |
float32 | [6] | 6 维动作信息:shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, gripper.pos |
episode_index |
int64 | [1] | 片段索引 |
frame_index |
int64 | [1] | 帧索引 |
index |
int64 | [1] | 样本索引 |
observation.images.camera1 |
video | [720, 1280, 3] | 摄像头1图像,分辨率 1280×720,RGB 3 通道,AV1 编码,30 FPS |
observation.images.camera2 |
video | [720, 1280, 3] | 摄像头2图像,分辨率 1280×720,RGB 3 通道,AV1 编码,30 FPS |
observation.images.camera3 |
video | [480, 640, 3] | 摄像头3图像,分辨率 640×480,RGB 3 通道,AV1 编码,30 FPS |
observation.state |
float32 | [6] | 6 维状态信息:shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, gripper.pos |
task_index |
int64 | [1] | 任务索引 |
timestamp |
float32 | [1] | 时间戳 |
其他信息
- 总任务数:1
- 总片段数(episodes):100
- 代码库版本:v3.0
- 可视化:可通过 LeRobot 可视化工具 查看数据集。



