HSJUSER/ffw_sg2_rev1_Test_A5

该数据集是使用LeRobot工具创建的，专为机器人学任务设计，涉及机器人类型ffw_sg2_rev1。数据集包含11个episodes，总计6604帧，覆盖1个任务，并提供了33个视频。所有数据都用于训练分割，帧率为30fps。数据以parquet格式存储，并包括以下特征：时间戳、帧索引、episode索引、索引、任务索引，以及来自三个摄像头的图像观察（头部摄像头分辨率为376x672，左腕和右腕摄像头分辨率为240x424），状态观察包含19个关节状态（如左右臂和头部的关节），动作数据也包含19个关节动作。该数据集适用于机器人控制、模仿学习或强化学习研究，旨在支持机器人行为分析和算法开发。

This dataset was created using the LeRobot tool and is designed for robotics tasks, specifically involving the robot type ffw_sg2_rev1. It consists of 11 episodes, totaling 6604 frames, covering 1 task, and includes 33 videos. All data is allocated to the training split with a frame rate of 30fps. The data is stored in parquet format and features the following: timestamp, frame index, episode index, index, task index, image observations from three cameras (head camera with resolution 376x672, left and right wrist cameras with resolution 240x424), state observations comprising 19 joint states (such as joints for left and right arms and head), and action data also with 19 joint actions. The dataset is suitable for robotics control, imitation learning, or reinforcement learning research, aiming to support robot behavior analysis and algorithm development.