robotgeneralist/cs-tablecloth
收藏资源简介:
该数据集是一个机器人学数据集,使用LeRobot创建,专门用于机器人任务。数据集包含131个episodes,总计239219个frames,数据以parquet格式存储。数据集结构包括多种观察特征,如UR机器人的TCP位姿(6维浮点数组,表示位置和方向)、工具状态(1维浮点数组)、力矩数据(6维浮点数组,表示力和扭矩),以及来自三个不同视角的图像数据(侧视、左腕、右腕,均为视频格式,分辨率分别为300x355和360x270,50fps)。此外,还包括ileader设备的TCP位姿(7维浮点数组,含四元数方向)、按钮状态(4维浮点数组)和工具值(1维浮点数组)。动作特征包含对UR机器人的命令(7维浮点数组,控制位置、方向和工具)。元数据包括时间戳、帧索引、episode索引等。数据集仅用于训练,采用Apache 2.0许可证。
This dataset is a robotics dataset created using LeRobot, designed for robot tasks. It contains 131 episodes with a total of 239,219 frames, stored in parquet format. The dataset structure includes multiple observation features such as UR robot TCP pose (6-dimensional float array for position and orientation), tool state (1-dimensional float array), wrench data (6-dimensional float array for force and torque), and image data from three different perspectives (side, wrist_left, wrist_right, all in video format with resolutions of 300x355 and 360x270, 50fps). Additionally, it includes ileader device TCP pose (7-dimensional float array with quaternion orientation), button states (4-dimensional float array), and tool value (1-dimensional float array). Action features consist of commands for the UR robot (7-dimensional float array controlling position, orientation, and tool). Metadata includes timestamps, frame index, episode index, etc. The dataset is intended for training only and is licensed under Apache 2.0.




