遇见数据集

so101_pick_red_cube_tray_sim_50_v29_homehold_success

收藏
Hugging Face2026-06-14 更新2026-06-14 收录
官方服务:

资源简介:

该数据集是一个机器人学数据集,使用LeRobot工具创建,专注于模拟环境中的机器人任务。它包含50个episodes,总计88696帧,帧率为30fps。数据以Parquet文件格式存储,视频以MP4格式存储。特征包括:动作数据(6维浮点数组,表示机器人关节位置,包括肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置)、观测状态(6维浮点数组,与动作数据相同)、来自三个摄像头的图像观测(每个摄像头输出480x640分辨率的RGB视频,3个通道),以及时间戳、帧索引、episode索引、任务索引等元数据。数据集用于训练和评估机器人控制算法,具体任务涉及拾取红色立方体托盘,但未在描述中详细说明。数据集结构支持机器学习模型训练,适用于机器人动作预测和视觉感知任务。

This dataset is a robotics dataset created using the LeRobot tool, focusing on robot tasks in simulated environments. It contains 50 episodes, totaling 88,696 frames at a frame rate of 30fps. The data is stored in Parquet file format, with videos in MP4 format. Features include: action data (a 6-dimensional floating-point array representing robot joint positions, including shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist rotation, and gripper position), observation state (a 6-dimensional floating-point array, same as action data), image observations from three cameras (each camera outputs RGB video at 480x640 resolution with 3 channels), and metadata such as timestamps, frame indices, episode indices, and task indices. The dataset is used for training and evaluating robot control algorithms, with specific tasks involving picking up a red cube tray, though not detailed in the description. The dataset structure supports machine learning model training and is suitable for robot action prediction and visual perception tasks.

提供机构:
tatsuyaaaaaaa
创建时间:
2026-06-14
原始信息汇总

数据集概述

  • 数据集名称: so101_pick_red_cube_tray_sim_50_v29_homehold_success
  • 许可证: Apache-2.0
  • 任务类别: 机器人学(robotics)
  • 标签: LeRobot

数据集规模

  • 总片段数(episodes): 50
  • 总帧数: 88,696
  • 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
  • 数据集划分: 训练集(train)包含全部50个片段

数据集结构

  • 代码库版本: v3.0
  • 机器人类型: so_follower
  • 数据分块(chunk)大小: 1000
  • 数据文件路径: data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
  • 视频文件路径: videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4

特征说明

特征名称 数据类型 形状 说明
action float32 (6,) 动作:包含6个关节位置(肩部旋转、肩部升降、肘部弯曲、手腕弯曲、手腕旋转、夹爪)
observation.state float32 (6,) 观察状态:与动作相同的6个关节位置
observation.images.camera1 video (480, 640, 3) 摄像头1视频:分辨率480x640,H.264编码,30 FPS
observation.images.camera2 video (480, 640, 3) 摄像头2视频:分辨率480x640,H.264编码,30 FPS
observation.images.camera3 video (480, 640, 3) 摄像头3视频:分辨率480x640,H.264编码,30 FPS
timestamp float32 (1,) 时间戳
frame_index int64 (1,) 帧索引
episode_index int64 (1,) 片段索引
index int64 (1,) 索引
task_index int64 (1,) 任务索引
二维码
社区交流群
二维码
科研交流群
商业服务