libero_10_sawyer_additionalCams_failures

该数据集是一个基于LeRobot框架创建的机器人学习数据集，专门针对Sawyer机器人。它包含499个训练episodes，总计137580帧，覆盖10个不同任务。数据以每秒20帧的速率采集，提供多模态观测：状态观测（如8维状态向量、6维末端执行器状态、7维关节状态、2维夹爪状态）和图像观测（前视图和侧视图视频，分辨率为256x256，RGB格式）。此外，数据集还包括7维动作数据、时间戳、帧索引和episode索引。数据以parquet文件存储观测数据，mp4文件存储视频，总数据量约100MB（观测数据）和约200MB（视频），适用于机器人控制、强化学习等研究任务。This dataset is a robotics learning dataset developed based on the LeRobot framework, tailored specifically for the Sawyer robot. It contains 499 training episodes, amounting to a total of 137,580 frames across 10 distinct tasks. Data is collected at 20 frames per second, and offers multimodal observations: state observations including an 8-dimensional state vector, 6-dimensional end-effector state, 7-dimensional joint state, and 2-dimensional gripper state, as well as image observations consisting of front-view and side-view RGB videos with a resolution of 256×256. In addition, the dataset also provides 7-dimensional action data, timestamps, frame indices, and episode indices. The observation data is stored in Parquet files while videos are stored in MP4 files, with the total size of the dataset being approximately 100 MB for observation data and 200 MB for videos. This dataset is applicable to research tasks such as robot control and reinforcement learning.

原始信息汇总

数据集概述

• 数据集名称: libero_10_sawyer_additionalCams_failures
• 发布者: OliverHausdoerfer
• 许可证: Apache-2.0
• 任务类别: 机器人学 (Robotics)
• 标签: LeRobot, libero, sawyer

数据集规模与组成

• 总片段数（Episodes）: 499
• 总帧数（Frames）: 137,580
• 总任务数（Tasks）: 10
• 帧率（FPS）: 20
• 数据文件大小: 约 100 MB
• 视频文件大小: 约 200 MB

数据结构

数据存储格式

• 数据路径: data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet (Parquet 格式)
• 视频路径: videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4 (MP4 格式)
• 分块大小（Chunks size）: 1000 帧

特征（Features）

特征名称	数据类型	形状	说明
observation.state	float32	[8]	机器人状态：x, y, z, axis_angle1-3, gripper (×2)
observation.states.ee_state	float32	[6]	末端执行器状态：x, y, z, axis_angle1-3
observation.states.joint_state	float32	[7]	关节状态：joint_0 至 joint_6
observation.states.gripper_state	float32	[2]	夹爪状态：gripper (×2)
observation.images.frontview_image	video	[256, 256, 3]	前视图图像（AV1 编码，20 FPS）
observation.images.sideview_image	video	[256, 256, 3]	侧视图图像（AV1 编码，20 FPS）
action	float32	[7]	动作：x, y, z, axis_angle1-3, gripper
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

数据集划分

• 训练集: 索引 0 至 498（共 499 个片段）

机器人平台

• 机器人类型: Sawyer