test_vla_dataset

该数据集是一个专为机器人操作任务（尤其是拾放任务）设计的数据集，使用LeRobot工具创建。它集成了多模态传感器数据，包括来自顶部摄像头和夹爪摄像头的视频观察（分辨率为720x1280，帧率为30fps）、深度图像（分辨率为720x1280）、机器人关节状态（8维，涵盖6个关节和2个夹爪电机）以及末端执行器状态（位置和四元数姿态）。数据以Parquet格式存储，总大小为数据文件100MB和视频文件200MB，包含2个episodes和492帧，覆盖1个任务。该数据集适用于机器人学习研究领域，如模仿学习和强化学习。

This dataset is tailored for robotic manipulation tasks, particularly pick-and-place operations, and was constructed using the LeRobot toolkit. It integrates multimodal sensor data, including visual observations from both the overhead camera and the gripper camera (resolution: 720×1280, frame rate: 30fps), depth images (resolution: 720×1280), robot joint states (8-dimensional, covering 6 joints and 2 gripper motors), and end-effector states (position and quaternion pose). The data is stored in Parquet format, with a total size of 100MB for data files and 200MB for video files. It contains 2 episodes and 492 frames, covering 1 single task. This dataset is suitable for research in robotic learning, such as imitation learning and reinforcement learning.