tutorial_vla_dataset

该数据集使用LeRobot创建，是一个机器人交互数据集，包含50个交互片段和9758帧数据。数据集结构包括图像观察、状态、动作和物体姿态等特征，数据以parquet文件格式存储，并包含视频文件。具体特征包括观察图像（256x256x3）、手腕图像（256x256x3）、状态（8维浮点数）、动作（7维浮点数）、末端执行器姿态（7维浮点数）、物体姿态（10x6维浮点数）等。数据集还包含时间戳、帧索引、片段索引等元数据。This robotic interaction dataset was developed using LeRobot, comprising 50 interaction segments and 9758 frames of data. The dataset structure includes features such as image observations, states, actions, and object poses, with all data stored in Parquet file format and accompanied by video files. Specific features include: observation images (256×256×3), wrist images (256×256×3), 8-dimensional floating-point states, 7-dimensional floating-point actions, end-effector poses (7-dimensional floating-point values), and object poses (10×6-dimensional floating-point values), among others. The dataset also contains metadata such as timestamps, frame indices, and segment indices.