oxe-aug/language_table_train_180000_185000_augmented

language_table_train_180000_185000_augmented是一个机器人交互数据集，包含多种机器人在执行任务时的增强图像、位姿、关节角度等信息，以及自然语言指令。数据集适用于机器人策略学习领域，共有5000个剧集，78347帧图像，支持跨机器人模型的策略学习。

The language_table_train_180000_185000_augmented dataset is a part of the OXE-AugE project, which is a large-scale robot dataset augmented from the original OXE dataset. It includes various robots (google_robot, images, jaco, kinova3, kuka_iiwa, panda, sawyer, ur5e) interacting in specific tasks with augmented images, poses, joint angles, and other related information. Each video is split into multiple chunks stored in .parquet file format, accompanied by corresponding .mp4 video files. The dataset consists of 5,000 episodes with a total of 78,347 frames, covering augmented images, poses, joint angles of multiple robots, as well as natural language instructions and task indices. The dataset is used to support policy learning across different robot models.