ankile/square-01a-baseline-uniform-r2-policy-rollouts

该数据集是一个机器人操作数据集，使用LeRobot框架创建，专门针对Franka Emika Panda机器人。它包含100个训练episodes，总计21481帧数据，用于单一任务（可能涉及螺母操作，如抓取或装配）。数据集以parquet格式存储观测和动作数据，同时提供mp4格式的视频文件。关键特征包括：观测状态（如末端执行器的位置、四元数姿态和夹爪位置）、动作（如末端执行器的增量位置和旋转变化，以及夹爪动作）、环境状态（如螺母相对于末端执行器的位置和姿态）、奖励、成功标志、完成标志等。数据以20 FPS采集，总数据文件大小为100 MB，视频文件大小为200 MB。数据集结构支持机器人强化学习和模仿学习任务，适用于训练和评估机器人控制算法。This dataset is a robotic manipulation dataset created using the LeRobot framework, specifically for the Franka Emika Panda robot. It contains 100 training episodes with a total of 21,481 frames, designed for a single task (likely involving nut manipulation, such as grasping or assembly). The data is stored in parquet format for observations and actions, along with mp4 video files. Key features include: observation state (e.g., end-effector position, quaternion orientation, and gripper joint positions), action (e.g., incremental end-effector position and rotation changes, and gripper action), environment state (e.g., nut position and orientation relative to the end-effector), reward, success flag, done flag, and more. Data is collected at 20 FPS, with total data file size of 100 MB and video file size of 200 MB. The dataset structure supports robotic reinforcement learning and imitation learning tasks, suitable for training and evaluating robot control algorithms.