wrong_rollout_eval_act_2026-05-29_shakeitup_dragonfly_checkpoint_20_000

该数据集是一个机器人学数据集，使用LeRobot工具创建，专注于收集机器人动作和观察数据。它包含19个总episodes和15778个总frames，数据以parquet文件格式存储，帧率为30fps。特征包括动作（如肩部平移、肩部提升、肘部弯曲、手腕弯曲、手腕滚动和夹爪位置）、观察状态（关节位置）、图像数据（顶部和手腕摄像头的视频，分辨率为480x640，3通道）以及左触觉频谱图（分辨率为224x224，3通道）。此外，还包含时间戳、帧索引、episode索引、任务索引等元数据。数据集针对机器人类型so101_follower_dragontactile设计，并提供了训练分割（所有episodes用于训练）。数据总大小约为100MB（数据文件）和200MB（视频文件），适用于机器人控制、感知和机器学习研究。This robotics dataset was developed using the LeRobot toolkit, with a core focus on collecting robot motion and observation data. It contains 19 total episodes and 15778 total frames, with all data stored in Parquet file format at a frame rate of 30 fps. The dataset's features include actions (such as shoulder translation, shoulder lift, elbow flexion, wrist flexion, wrist roll, and gripper position), observation states (joint positions), image data (videos captured by the top and wrist cameras with a resolution of 480×640 and 3 color channels), and left tactile spectrograms (with a resolution of 224×224 and 3 color channels). Additionally, it includes metadata such as timestamps, frame indices, episode indices, and task indices. This dataset is designed for the robot platform so101_follower_dragontactile, and a training split is provided where all episodes are used for training. The total size of the dataset is approximately 100 MB for data files and 200 MB for video files, making it suitable for research in robot control, perception, and machine learning.

原始信息汇总

数据集概述

• 数据集名称: wrong_rollout_eval_act_2026-05-29_shakeitup_dragonfly_checkpoint_20_000
• 许可证: Apache-2.0
• 任务类别: 机器人技术 (Robotics)
• 创建工具: 使用 LeRobot 创建
• 可视化: 可通过 可视化数据集 交互查看

数据集结构

• 代码库版本: v3.0
• 帧率 (FPS): 30
• 总集数 (episodes): 19
• 总帧数 (frames): 15,778
• 总任务数 (tasks): 1
• 数据类型: Parquet 文件 (动作、状态、时间戳等) + 视频文件 (观察图像)
• 数据文件大小: ~100 MB
• 视频文件大小: ~200 MB
• 机器人类型: so101_follower_dragontactile
• 数据划分: 训练集 (train) 包含全部 19 集 (0-18)

特征 (Features)

特征名称	数据类型	形状	描述
action	float32	[6]	机器人动作指令 (6个关节: shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper)
observation.state	float32	[6]	机器人状态 (与动作相同的6个关节位置)
observation.images.top	video	[480, 640, 3]	顶部摄像头图像 (AV1编码、30 FPS)
observation.images.wrist	video	[480, 640, 3]	腕部摄像头图像 (AV1编码、30 FPS)
observation.images.left_tactile_spectrogram	video	[224, 224, 3]	左侧触觉传感器频谱图 (AV1编码、30 FPS)
timestamp	float32	[1]	时间戳
frame_index	int64	[1]	帧索引
episode_index	int64	[1]	集索引
index	int64	[1]	全局索引
task_index	int64	[1]	任务索引

引用信息

• 论文: 暂无提供
• 主页: 暂无提供