move_clutter_2
收藏资源简介:
该数据集是使用LeRobot项目创建的机器人学数据集,适用于机器人学习任务。数据集采用Apache-2.0许可证,包含106个episodes,总计62,984帧数据,覆盖4个不同的任务。数据以Parquet文件格式存储,采集帧率为30fps。数据集包含多模态信息:动作数据为11维浮点向量,包括末端执行器速度、手腕速度、手指速度、接触向量及episode结束标志;观测状态为44维浮点向量,涵盖夹爪的位置、旋转、速度、传感器读数(如激光测距、压力、张力)以及多个目标物体的方位和距离信息;视觉观测包含来自四个摄像头的视频流:夹爪相机(分辨率384x384)、俯视相机(分辨率512x512)和两个锚定相机(分辨率均为544x960),所有视频均为30fps的RGB格式,使用AV1编码。此外,数据集还包含时间戳、帧索引、episode索引和任务索引等元数据。数据总大小约为300MB(其中数据文件100MB,视频文件200MB)。该数据集适用于机器人模仿学习、强化学习、视觉运动控制等研究场景。
This is a robotics dataset developed with the LeRobot project, tailored for robotic learning tasks. The dataset is licensed under Apache-2.0, containing 106 episodes with a total of 62,984 frames, covering 4 distinct tasks. The data is stored in Parquet file format, with an acquisition frame rate of 30 fps. The dataset includes multimodal information: the action data is an 11-dimensional floating-point vector, encompassing end-effector velocity, wrist velocity, finger velocity, contact vector, and episode termination flag; the observation state is a 44-dimensional floating-point vector, covering the position, rotation, velocity of the gripper, sensor readings such as laser rangefinder data, pressure, and tension, as well as the orientation and distance information of multiple target objects; the visual observations include video streams from four cameras: the gripper camera (resolution 384×384), the top-down camera (resolution 512×512), and two anchored cameras (both with a resolution of 544×960), all videos are in RGB format at 30 fps and encoded with AV1. In addition, the dataset also contains metadata such as timestamps, frame indices, episode indices, and task indices. The total size of the dataset is approximately 300 MB, including 100 MB of data files and 200 MB of video files. This dataset is suitable for research scenarios such as robotic imitation learning, reinforcement learning, and visual motor control.
数据集概述
数据集名称: move_clutter_2
数据集地址: https://huggingface.co/datasets/naavox/move_clutter_2
许可协议: Apache-2.0
任务类别: 机器人学 (Robotics)
创建工具: 使用 LeRobot 创建
数据集结构
基本信息
- 代码库版本: v3.0
- 帧率 (FPS): 30
- 总片段数 (Episodes): 137
- 总帧数 (Frames): 86847
- 总任务数 (Tasks): 6
- 数据文件大小: 100 MB
- 视频文件大小: 200 MB
- 机器人类型: stringman
- 数据集划分: 训练集 (train): 0-137 片段
特征 (Features)
动作 (Action)
- 数据类型: float32
- 形状: [11]
- 特征名称: vel_x, vel_y, vel_z, room_vel_x, room_vel_y, wrist_speed, finger_speed, contact_vec_x, contact_vec_y, contact_vec_z, episode_end
观测状态 (observation.state)
- 数据类型: float32
- 形状: [44]
- 特征名称: 包括速度、位置、旋转、角度、压力、轴承、距离、张力、视觉位置等44个变量
观测图像 (observation.images)
数据集包含4个相机视角的图像观测,均为视频格式:
| 相机名称 | 分辨率 (H x W x C) | 编码格式 | 帧率 | 颜色通道 |
|---|---|---|---|---|
| gripper_camera | 384 x 384 x 3 | AV1 | 30 | RGB |
| overhead_camera | 512 x 512 x 3 | AV1 | 30 | RGB |
| anchor_camera_0 | 544 x 960 x 3 | AV1 | 30 | RGB |
| anchor_camera_1 | 544 x 960 x 3 | AV1 | 30 | RGB |
其他特征
| 特征名 | 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] | 全局索引 |
| task_index | int64 | [1] | 任务索引 |
数据存储与路径
- 数据文件:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet(Parquet 格式) - 视频文件:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4(MP4 格式) - 配置文件: meta/info.json




