full_featured
收藏Hugging Face2026-05-31 更新2026-05-28 收录
下载链接:
https://huggingface.co/datasets/naavox/full_featured
下载链接
链接失效反馈官方服务:
资源简介:
该数据集使用LeRobot工具创建,是一个机器人学数据集,专门用于机器人控制与学习任务。数据集包含22个episodes,总计25654帧,覆盖10个不同任务。数据以parquet格式存储,视频以mp4格式存储,帧率为30fps。数据集的特征包括动作(包含5个维度:vel_x、vel_y、vel_z、wrist_speed、finger_speed)、观测状态(包含43个维度,如机器人位置、速度、传感器数据等)、图像观测(来自三个摄像头:gripper_camera(分辨率384x384)、anchor_camera_1和anchor_camera_2(分辨率均为544x960))、时间戳、帧索引、episode索引、索引和任务索引。机器人类型为stringman,数据集适用于机器人控制、强化学习和计算机视觉研究。
This dataset was developed using the LeRobot toolkit, and it is a robotics dataset specifically tailored for robot control and learning tasks. It consists of 22 episodes, totaling 25,654 frames across 10 distinct tasks. The dataset is stored in Parquet format, while the accompanying videos are saved in MP4 format at a frame rate of 30 fps. The dataset features include actions with 5 dimensions: vel_x, vel_y, vel_z, wrist_speed, and finger_speed; observation states with 43 dimensions such as robot position, velocity, and sensor data; image observations from three cameras: gripper_camera with a resolution of 384×384, anchor_camera_1 and anchor_camera_2 both with a resolution of 544×960; timestamps, frame indices, episode indices, indices, and task indices. The robot type for this dataset is stringman, and it is applicable for research in robot control, reinforcement learning, and computer vision.
提供机构:
naavox创建时间:
2026-05-28
原始信息汇总
数据集概述
- 数据集名称: full_featured
- 数据集地址: https://huggingface.co/datasets/naavox/full_featured
- 许可证: Apache-2.0
- 任务类别: 机器人学(robotics)
- 创建工具: 基于 LeRobot 创建
数据集规模
| 指标 | 数值 |
|---|---|
| 总剧集数(episodes) | 22 |
| 总帧数(frames) | 25654 |
| 总任务数(tasks) | 10 |
| 数据文件大小(parquet) | 约 100 MB |
| 视频文件大小(mp4) | 约 200 MB |
| 帧率(fps) | 30 |
| 训练集划分 | 全部剧集(0:22) |
机器人类型
- 机器人类型:
stringman
特征结构
动作(action)
- 数据类型: float32
- 维度: 5
- 各分量含义: vel_x, vel_y, vel_z, wrist_speed, finger_speed
观察-状态(observation.state)
- 数据类型: float32
- 维度: 43
- 包含信息: 速度、夹爪位置与旋转、手指角度、激光测距、压力、手腕角度、目标力、物体方位角与距离、摆动取消标志、张力、龙门位置、视觉位置、悬挂位置等
观察-图像(observation.images)
共包含三个摄像头视角:
- gripper_camera(夹爪摄像头)
- 分辨率:384×384
- 色彩通道:3(RGB)
- 视频编码:AV1
- 帧率:30 fps
- anchor_camera_1(锚定摄像头1)
- 分辨率:544×960
- 色彩通道:3(RGB)
- 视频编码:AV1
- 帧率:30 fps
- anchor_camera_2(锚定摄像头2)
- 分辨率:544×960
- 色彩通道:3(RGB)
- 视频编码:AV1
- 帧率:30 fps
时间戳与索引
- timestamp: float32,形状 [1]
- frame_index: int64,形状 [1]
- episode_index: int64,形状 [1]
- index: int64,形状 [1]
- task_index: int64,形状 [1]
数据文件结构
- 数据路径格式:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频路径格式:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4 - 分块大小: 1000 帧/块
引用信息
- BibTeX: 待补充



