oxe-auge/fractal20220817_data_train_15000_17500_augmented

fractal20220817_data_train_15000_17500_augmented数据集是OXE-AugE项目的一部分，专注于机器人技术的增强数据。数据集包含多种机器人的视频帧、机器人姿态、关节位置和自然语言指令。具体包括：代码库版本v3.0，支持的机器人有images、jaco、kinova3、kuka_iiwa、panda、sawyer、ur5e和xarm7，帧率为3FPS，总共有2,500个片段和108,904帧。数据集结构以parquet文件格式存储，包含训练集（0:2500）。特征包括机器人的图像、末端执行器姿态、关节角度、自然语言指令等。该数据集旨在用于训练跨体现策略的机器人学习任务。

The fractal20220817_data_train_15000_17500_augmented dataset is part of the OXE-AugE project, focusing on augmented data for robotics. It includes video frames, robot poses, joint positions, and natural language instructions for various robots. Details include: codebase version v3.0, supported robots (images, jaco, kinova3, kuka_iiwa, panda, sawyer, ur5e, xarm7), frame rate of 3FPS, total of 2,500 episodes and 108,904 frames. The dataset is structured in parquet files, with a train split (0:2500). Features include robot images, end-effector poses, joint angles, natural language instructions, etc. The dataset is designed for training cross-embodiment policy learning in robotics.