franka_pnp

这是一个由LeRobot创建的机器人任务数据集，专为机器人学习和控制任务设计。数据集包含50个训练片段，总帧数为12477，帧率为20fps，使用Franka机器人平台。数据以Parquet格式存储，包括观察图像（如主图像和手腕图像，尺寸为256x256x3）、机器人状态（6维浮点数）、动作（8维浮点数）、对象初始状态（6维浮点数）以及时间戳、帧索引、片段索引等元数据。数据集结构详细，适用于机器人强化学习或模仿学习任务，许可证为Apache-2.0。

This is a robot task dataset created by LeRobot, specifically designed for robot learning and control tasks. The dataset contains 50 training segments with a total of 12477 frames, a frame rate of 20fps, and uses the Franka robot platform. Data is stored in Parquet format, including observation images (such as main and wrist images, sized 256x256x3), robot states (6-dimensional floats), actions (8-dimensional floats), object initial states (6-dimensional floats), as well as metadata like timestamps, frame indices, and segment indices. The dataset is structured in detail, suitable for robot reinforcement learning or imitation learning tasks, with an Apache-2.0 license.