fecasado/blueberry-sandwich-320x240-filtered

该数据集是一个机器人控制数据集，使用LeRobot框架创建，专为机器人学任务设计。数据集包含264个训练episodes，总计210,438帧，帧率为15fps。数据特征包括动作向量（26维，控制左右机械臂的线性和角速度、手指关节以及基座移动）、观察状态（29维，包括左右机械臂和手部关节位置、视线坐标及有效性）、以及多视角图像数据（左、右、用户和用户视线视角的视频，分辨率240x320，3通道，AV1编码）。此外，数据集提供时间戳、帧索引、episode索引、任务索引等元信息。机器人类型为blueberry_ros，数据分割为训练集，涵盖4个不同任务。数据文件以parquet格式存储，总大小约300MB（数据100MB，视频200MB）。

This dataset is a robotics control dataset created using the LeRobot framework, designed for robotics tasks. It contains 264 training episodes, totaling 210,438 frames at 15 fps. Features include action vectors (26-dimensional, controlling linear and angular velocities of left and right arms, finger joints, and base movement), observation states (29-dimensional, including joint positions of left and right arms and hands, gaze coordinates, and validity), and multi-view image data (videos from left, right, user, and user gaze perspectives, with resolution 240x320, 3 channels, AV1 encoded). Additionally, the dataset provides metadata such as timestamps, frame indices, episode indices, and task indices. The robot type is blueberry_ros, with data split into training set covering 4 different tasks. Data files are stored in parquet format, with a total size of approximately 300MB (100MB for data, 200MB for videos).