pick-up-green-cube-and-place-on-the-right-side-o-20260723-180358-iter1-merged

该数据集是一个机器人演示数据集，使用LeRobot工具创建，记录了机器人执行拾取绿色立方体并放置到右侧任务的62个完整演示片段。数据集包含总共44807帧数据，以30帧每秒的速率采集。每个数据点包括机器人的6维关节位置动作（如肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转、夹爪位置）、对应的6维关节位置观察状态、以及来自腕部（480x640像素）、侧面（480x848像素）和顶部（480x848像素）三个视角的RGB视频观察。此外，还包含时间戳、帧索引、片段索引、任务索引等元数据。数据集用于机器人学习任务，如模仿学习或强化学习，支持训练机器人执行类似操作任务。数据集结构为Parquet格式，总数据文件大小为100MB，视频文件大小为200MB。This dataset is a robotic demonstration dataset created using the LeRobot tool, which contains 62 complete demonstration episodes for the task of a robot picking up a green cube and placing it to the right. The dataset consists of a total of 44,807 frames collected at a rate of 30 frames per second. Each data point includes the robot's 6-dimensional joint position actions (such as shoulder translation, shoulder lift, elbow flexion, wrist flexion, wrist rotation, and gripper position), the corresponding 6-dimensional joint position observation states, as well as RGB video observations from three perspectives: wrist-mounted (480x640 pixels), side-view (480x848 pixels), and top-view (480x848 pixels). In addition, it includes metadata such as timestamps, frame indices, episode indices, and task indices. This dataset is intended for robotic learning tasks such as imitation learning or reinforcement learning, supporting the training of robots to perform similar manipulation tasks. The dataset is structured in Parquet format, with a total data file size of 100 MB and a total video file size of 200 MB.

原始信息汇总

数据集概述

• 数据集名称: pick-up-green-cube-and-place-on-the-right-side-o-20260723-180358-iter1-merged
• 许可证: Apache-2.0
• 任务类别: 机器人学 (Robotics)
• 标签: LeRobot
• 创建工具: 使用 LeRobot 创建

数据集统计

• 总片段数 (Episodes): 62
• 总帧数 (Frames): 44,807
• 总任务数: 1
• 帧率 (FPS): 30
• 数据文件大小: 100 MB
• 视频文件大小: 200 MB
• 机器人类型: so_follower
• 数据划分: 训练集 (train) 包含所有 62 个片段

特征结构

动作 (action)

• 数据类型: float32
• 形状: [6]
• 名称: shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, gripper.pos

观测状态 (observation.state)

• 数据类型: float32
• 形状: [6]
• 名称: shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, gripper.pos

观测图像

腕部相机 (observation.images.wrist)

• 数据类型: video
• 形状: [480, 640, 3] (高度, 宽度, 通道)
• 编码: av1, 30 FPS, yuv420p, 无音频

侧方相机 (observation.images.side)

• 数据类型: video
• 形状: [480, 848, 3] (高度, 宽度, 通道)
• 编码: av1, 30 FPS, yuv420p, 无音频

顶部相机 (observation.images.top)

• 数据类型: video
• 形状: [480, 848, 3] (高度, 宽度, 通道)
• 编码: av1, 30 FPS, yuv420p, 无音频

其他特征

• 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]

数据文件组织

• 数据路径: data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
• 视频路径: videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
• 块大小 (chunks_size): 1000

可视化

• 可通过 可视化数据集 在线查看数据。