Jiafei1224/09042026-cup-stacking-combined

该数据集是一个机器人学习数据集，使用LeRobot工具创建，专注于机器人控制任务。数据集包含200个episodes，总计163903帧，涉及一个任务（可能为堆叠杯子任务，基于数据集名称推断）。数据以块形式存储，每个块1000帧，数据文件大小为100MB，视频文件大小为200MB，帧率为30fps。数据集仅包含训练集。特征包括：动作数据（14维浮点数组，表示左右机械臂的6个关节位置和夹爪位置）、观察状态数据（与动作数据相同的14维浮点数组，反映机器人状态）、来自三个摄像头（右、左、顶部）的观察图像（视频格式，分辨率360x640，3通道RGB，30fps，使用AV1编解码器），以及辅助信息如时间戳、帧索引、episode索引、索引和任务索引。机器人类型为bi_yam_follower，可能表示双机械臂系统。数据集适用于机器人强化学习或模仿学习研究，用于训练和评估控制算法。This dataset is a robotics learning dataset created using the LeRobot tool, focusing on robot control tasks. It contains 200 episodes with a total of 163,903 frames, involving one task (likely cup stacking, inferred from the dataset name). The data is stored in chunks of 1000 frames each, with data file size of 100 MB and video file size of 200 MB, at a frame rate of 30 fps. The dataset includes only a training set. Features include: action data (a 14-dimensional float32 array representing positions of 6 joints and gripper for both left and right robot arms), observation state data (a similar 14-dimensional float32 array reflecting robot state), observation images from three cameras (right, left, top) in video format (resolution 360x640, 3 channels RGB, 30 fps, using AV1 codec), and auxiliary information such as timestamp, frame index, episode index, index, and task index. The robot type is bi_yam_follower, possibly indicating a dual-arm system. The dataset is suitable for robotics reinforcement learning or imitation learning research, for training and evaluating control algorithms.