Visual monitoring of civil infrastructure systems via camera-equipped Unmanned Aerial Vehicles (UAVs): a review of related works

综述搭载相机的无人机对施工现场和既有基础设施的监测，分为数据采集、分析和可视化三个阶段。采集涵盖航线规划与自主飞行，分析以基于图像的三维重建（结合或不结合 BIM）为核心，可视化涵盖 AR 和网页点云工具。文章指出当时的实践仍依赖熟练飞手和 GPS 航点，并列举了实现自主飞行的具体障碍，如 GPS 信号丢失和钢结构附近的磁力计问题。

主要发现

• Over a thousand images may be collected per flight on a construction site
• Resulting point clouds can exceed 100 million points
• Processing needs to finish within a few hours to support project control
• GPS-based flight planning ignores site dynamics and suffers from GPS loss and magnetometer issues near steel
• SLAM had rarely been tested on civil infrastructure at the time

局限

• Literature up to 2015; autonomy and AI have advanced since
• Tables of applications summarise rather than benchmark methods