oxe-auge/fractal20220817_data_train_37500_40000_augmented

fractal20220817_data_train_37500_40000_augmented数据集是OXE-AugE项目的一部分，专注于机器人技术的跨体现策略学习。该数据集包含多种机器人的增强图像和运动学数据，以及自然语言指令和情节元数据。数据集结构包括2500个情节，共109327帧，帧率为3fps。数据以parquet文件格式存储，包含机器人的图像、关节位置、末端执行器姿态等特征。数据集采用CC-BY-4.0许可，要求在使用时同时引用本数据集和原始上游数据集。

The fractal20220817_data_train_37500_40000_augmented dataset is part of the OXE-AugE project, focusing on cross-embodiment policy learning in robotics. It includes augmented images and kinematic data for various robots, along with natural language instructions and episode metadata. The dataset consists of 2,500 episodes with a total of 109,327 frames at 3fps. Data is stored in parquet files and includes features such as robot images, joint positions, and end-effector poses. Licensed under CC-BY-4.0, it requires citation of both this dataset and the original upstream datasets when used.