oxe-aug/language_table_train_15000_20000_augmented

language_table_train_15000_20000_augmented是一个机器人增强数据集，包含多种机器人的增强图像、关节位置等信息。数据集基于oxe_aug代码库的v3.0版本，共有5000个剧集，79558帧，分为训练集。数据集中的机器人类型包括google_robot, images, jaco, kinova3, kuka_iiwa, panda, sawyer, ur5e等。每一帧都提供了丰富的机器人状态信息，包括末端执行器的位置、误差、关节角度等，并且还包括了描述任务的自然语言指令。

language_table_train_15000_20000_augmented is an augmented robot dataset containing enhanced images and joint positions for various robots. Based on the oxe_aug codebase version v3.0, the dataset includes 5,000 episodes and 79,558 frames, specifically distributed in a training set split. It covers robot types such as google_robot, images, jaco, kinova3, kuka_iiwa, panda, sawyer, ur5e, etc. Each frame provides rich robot state information, including the end effectors position, error, joint angles, and also includes natural language instructions describing the task.