eval_results

该数据集是使用LeRobot创建的机器人学习数据集，专门用于‘so_follower’机器人类型。数据集包含动作和观测状态特征，动作包括6个关节位置（如肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部滚动和夹爪位置），观测状态同样包含这些关节位置。此外，数据集提供来自三个摄像头的图像观测，每个摄像头分辨率为480x640像素、3个通道（RGB），以及时间戳、帧索引、剧集索引、索引和任务索引等元数据。数据以Parquet文件格式存储，视频文件以MP4格式存储，帧率为30fps。数据集总大小约为300MB（数据文件100MB，视频文件200MB），但剧集、帧和任务总数在元数据中显示为0，可能表示未指定或动态生成。许可证为Apache 2.0。This dataset is a robotic learning dataset developed using LeRobot, specifically tailored for the 'so_follower' robot type. It comprises action and observation state features. The actions consist of 6 joint positions, namely shoulder translation, shoulder lift, elbow flexion, wrist flexion, wrist roll, and gripper position, while the observation states also include these joint positions. Additionally, the dataset provides image observations from three cameras, each with a resolution of 480×640 pixels and 3 RGB channels, along with metadata including timestamps, frame indices, episode indices, individual data indices, and task indices. The data is stored in Parquet file format, whereas the video files are stored in MP4 format with a frame rate of 30 fps. The total size of the dataset is approximately 300 MB (100 MB for data files and 200 MB for video files). However, the total counts of episodes, frames, and tasks indicated in the metadata are 0, which may imply that these values are unspecified or dynamically generated. The license for this dataset is Apache 2.0.

原始信息汇总

数据集概述

• 数据集名称: zznzhang/eval_results
• 许可证: Apache-2.0
• 任务类别: 机器人学 (robotics)
• 标签: LeRobot
• 创建方式: 使用 LeRobot 框架生成

数据集结构

• 配置文件: default，数据文件路径为 data/*/*.parquet
• 元数据文件: meta/info.json
• 代码库版本: v3.0
• 机器人类型: so_follower
• 总剧集数: 0
• 总帧数: 0
• 总任务数: 0
• 块大小: 1000
• 数据文件大小: 100 MB
• 视频文件大小: 200 MB
• 帧率: 30 FPS
• 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

数据特征

特征名称	数据类型	形状	描述
action	float32	[6]	包含6个关节（shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper）的位置信息
observation.state	float32	[6]	观测到的状态，同样包含6个关节的位置信息
observation.images.camera1	video	[480, 640, 3]	来自摄像头的图像，分辨率480x640，3个通道
observation.images.camera2	video	[480, 640, 3]	来自摄像头的图像，分辨率480x640，3个通道
observation.images.camera3	video	[480, 640, 3]	来自摄像头的图像，分辨率480x640，3个通道
timestamp	float32	[1]	时间戳
frame_index	int64	[1]	帧索引
episode_index	int64	[1]	剧集索引
index	int64	[1]	索引
task_index	int64	[1]	任务索引