KS325/open-lower-drawer-r2-emp-train-expand

该数据集是一个用于机器人学习的专有数据集，专注于执行“打开下层抽屉”任务（任务名称：open-lower-drawer-r2-emp-train）。它使用LeRobot框架创建，包含28个训练episodes，总计20502个frames，数据以parquet文件格式存储。数据集结构丰富，特征包括：动作（action），包含6个浮点型关节位置（如肩部平移、肩部升降、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置）；观测（observation），包括机器人状态（6个关节位置）和两个摄像头图像（camera1和camera2），每个图像分辨率为480x640像素、3通道彩色视频，帧率为30fps；以及元数据如时间戳、帧索引、episode索引等。数据集适用于机器人控制、强化学习或计算机视觉任务，旨在模拟真实世界机器人操作场景。

This dataset is a specialized resource for robotics learning, focusing on the task of opening the lower drawer (task name: open-lower-drawer-r2-emp-train). It was created using the LeRobot framework and includes 28 training episodes with a total of 20502 frames, stored in parquet file format. The dataset features a comprehensive structure: actions (action) comprising 6 floating-point joint positions (e.g., shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, gripper position); observations (observation) including robot state (6 joint positions) and two camera images (camera1 and camera2), each with a resolution of 480x640 pixels, 3-channel color video at 30fps; along with metadata such as timestamp, frame index, and episode index. It is suitable for robotics control, reinforcement learning, or computer vision applications, designed to simulate real-world robot manipulation scenarios.