tingtu0721/record-dataset-200-2cameras

该数据集是一个机器人数据集，使用LeRobot工具创建，专注于so_follower机器人类型。数据集包含200个episodes，总计115,425帧，数据以parquet文件格式存储，视频以mp4格式存储，帧率为30fps。数据特征包括：动作数据（6个浮点型关节位置，对应肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置）、观测状态（与动作相同的6个关节位置）、观测图像（包括前视和顶视两个视角的RGB视频，分辨率均为480x640，3通道），以及时间戳、帧索引、episode索引等元数据。数据集用于机器人相关任务，如运动控制或视觉-动作学习。This dataset is a robotics dataset created using the LeRobot tool, focusing on the so_follower robot type. It contains 200 episodes with a total of 115,425 frames, stored in parquet files for data and mp4 files for videos, with a frame rate of 30fps. The features include: action data (6 floating-point joint positions corresponding to shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper positions), observation state (the same 6 joint positions), observation images (including front and top view RGB videos, both with 480x640 resolution and 3 channels), and metadata such as timestamp, frame index, and episode index. The dataset is intended for robotics-related tasks, such as motion control or vision-action learning.