yinongh/real_world_1_force_image
收藏资源简介:
该数据集使用LeRobot创建,专注于机器人学任务,具体涉及插头插入插座的模拟操作。数据集包含5个episodes,总计171帧和15个视频,采用30fps的帧率。数据结构包括动作特征(如delta_x、delta_y等9维浮点数)、观测特征(如末端执行器内力、姿态、对齐的插座深度图像、初始插头深度图像、点云数据、相机外参、手腕相机深度和RGB图像、辅助相机图像等),以及时间戳和索引信息。这些数据用于训练和评估机器人控制模型,特别是基于视觉和力反馈的精细操作任务。
This dataset was created using LeRobot and focuses on robotics tasks, specifically simulating plug insertion into a socket. It contains 5 episodes, with a total of 171 frames and 15 videos, recorded at 30fps. The dataset structure includes action features (e.g., delta_x, delta_y, etc., in 9-dimensional float32), observation features (such as end-effector internal forces, pose, aligned socket depth images, initial plug depth images, point cloud data, camera extrinsics, wrist camera depth and RGB images, auxiliary camera images, etc.), along with timestamps and indices. These data are intended for training and evaluating robot control models, particularly for fine-grained manipulation tasks based on visual and force feedback.




