kbot_cappuccino_count

该数据集是一个用于机器人任务的数据集，针对kbot_right_arm_follower机器人类型，专注于抓取放置（pick-and-place）任务，可能涉及制作卡布奇诺的场景。数据集包含224个episodes，总计166,988帧，以30帧每秒（fps）录制。数据以parquet文件格式存储，总数据文件大小为100MB，视频文件大小为200MB。数据集特征包括观测状态（如pitch、roll、yaw、elbow、wrist和gripper的位置）、来自三个摄像头视角（手腕、头顶和顶部）的视频图像（分辨率为720x1280，3通道），以及相应的动作数据。此外，还包含时间戳、帧索引、episode索引等元数据。数据集使用Apache 2.0许可证，并通过LeRobot工具创建。This dataset is designed for robotic tasks, specifically for the kbot_right_arm_follower robot type, focusing on pick-and-place tasks, potentially involving cappuccino-making scenarios. It contains 224 episodes with a total of 166,988 frames recorded at 30 frames per second (fps). The data is stored in parquet format, with a total data file size of 100MB and video file size of 200MB. Features include observation states (e.g., positions of pitch, roll, yaw, elbow, wrist, and gripper), video images from three camera views (wrist, overhead, and top) at 720x1280 resolution with 3 channels, and corresponding action data. Additionally, metadata such as timestamps, frame indices, and episode indices are included. The dataset uses the Apache 2.0 license and was created using the LeRobot tool.

原始信息汇总

数据集概述：infoslack/kbot_cappuccino_count

• 许可证: Apache-2.0
• 任务类别: 机器人技术 (robotics)
• 标签: LeRobot, kbot, robstride, smolvla, pick-and-place, cappuccino

数据集描述

该数据集使用 LeRobot 创建，专注于机器人抓取和放置 (pick-and-place) 任务，具体涉及咖啡杯（cappuccino）计数场景。

数据集结构

• 机器人类型: kbot_right_arm_follower
• 总片段数: 224
• 总帧数: 166,988
• 总任务数: 1
• 块大小: 1000
• 数据文件大小: 100 MB
• 视频文件大小: 200 MB
• 帧率: 30 FPS
• 数据集划分:
  • 训练集 (train): 0-224 (全部数据)

特征详情

• observation.state (观测状态):
  • 数据类型: float32
  • 形状: [6]
  • 维度标签: pitch.pos, roll.pos, yaw.pos, elbow.pos, wrist.pos, gripper.pos
• observation.images.wrist (腕部摄像头图像):
  • 数据类型: video
  • 形状: 720 x 1280 x 3 (高度 x 宽度 x 通道)
  • 编码: AV1, yuv420p, 30 FPS, 非深度图, 无音频
• observation.images.overhead (俯视摄像头图像):
  • 数据类型: video
  • 形状: 720 x 1280 x 3
  • 编码: 同 wrist 摄像头
• observation.images.top (顶部摄像头图像):
  • 数据类型: video
  • 形状: 720 x 1280 x 3
  • 编码: 同 wrist 摄像头
• action (动作):
  • 数据类型: float32
  • 形状: [6]
  • 维度标签: pitch.pos, roll.pos, yaw.pos, elbow.pos, wrist.pos, gripper.pos
• timestamp (时间戳): float32, 形状 [1]
• frame_index (帧索引): int64, 形状 [1]
• episode_index (片段索引): int64, 形状 [1]
• index (索引): int64, 形状 [1]
• task_index (任务索引): int64, 形状 [1]

数据路径

• 数据文件: data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
• 视频文件: videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4