robotgeneralist/cs-tablecloth-defected

该数据集是一个机器人学习数据集，使用LeRobot工具创建，专注于机器人操作任务。数据集包含181个episodes，总计360658帧数据，以50fps的帧率采集。数据存储为parquet格式，总数据文件大小为100MB，视频文件大小为200MB。数据集特征包括机器人UR10e的TCP姿态、工具状态、力/力矩传感器数据，以及来自侧面、左腕和右腕摄像头的多视角视频观测（分辨率分别为300x355和360x270）。此外，还包含领导者设备的TCP姿态、按钮状态和工具值，以及发送给UR机器人的动作命令（包括位置、姿态和工具控制）。数据集用于训练和评估机器人策略，支持模仿学习或强化学习任务。This is a robotics learning dataset developed with the LeRobot toolkit, dedicated to robotic manipulation tasks. The dataset consists of 181 episodes, totaling 360,658 frames collected at a frame rate of 50 fps. The data is stored in Parquet format, with an overall data file size of 100 MB and video file size of 200 MB. Dataset features include the TCP (Tool Center Point) pose of the UR10e robot, tool status, force/torque sensor data, as well as multi-view video observations captured by side, left wrist, and right wrist cameras (with resolutions of 300x355 and 360x270 respectively). Additionally, it contains the TCP pose of the leader device, button states and tool values, along with action commands sent to the UR robot, including position, pose and tool control. This dataset is intended for training and evaluating robotic policies, supporting tasks such as imitation learning or reinforcement learning.