saipuneethgottam/pickplace_merged_3cam_20260526

该数据集是使用LeRobot创建的机器人数据集，专为机器人学习任务设计。数据集包含1000个训练episodes，总计531,780帧数据，对应1个任务类型。数据存储为parquet格式，视频文件为mp4格式，帧率为30fps。数据集的特征包括：6维的机器人动作和状态（分别对应肩部平移、肩部升降、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置），以及来自三个摄像头的视觉观测（其中两个摄像头分辨率为1280x720，一个为640x480）。此外，还包含时间戳、帧索引、episode索引、任务索引等元数据。数据集主要用于机器人操作任务的模仿学习或强化学习研究。

This dataset is a robotics dataset created using LeRobot, designed for robot learning tasks. It contains 1000 training episodes, totaling 531,780 frames, corresponding to 1 task type. The data is stored in parquet format, with video files in mp4 format at 30 fps. The dataset features include: 6-dimensional robot actions and states (corresponding to shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper positions), as well as visual observations from three cameras (two with 1280x720 resolution and one with 640x480 resolution). Additionally, it includes metadata such as timestamps, frame indices, episode indices, and task indices. The dataset is primarily intended for imitation learning or reinforcement learning research in robot manipulation tasks.