Autonomous 3D Scanning of Construction Sites Using Terrestrial Laser Scanner and UGV

Accurate representation of construction sites for Artificial Intelligence (AI) processing and process monitoring demands complete 3D point clouds. Conventionally, a tripod-mounted terrestrial laser scanner (TLS) is employed for generating 3D point clouds at multiple locations, followed by post-processing, which is time-consuming. This paper proposes an autonomous scanning system integrating TLS with an Unmanned Ground Vehicle (UGV) to generate 3D point clouds. The approach employs frontier-based exploration to autonomously navigate the robot to optimal scan waypoints within predefined GPS-marked areas on an outdoor construction site. Subsequently, the high accuracy and long-range TLS scanner record a point cloud, which is registered, and a new optimal waypoint for the robot is computed. This iterative exploration and scanning process continues until the point cloud is complete and no new scanning frontier is identified within the scan area. The effectiveness of the newly developed autonomous scanning technique is evaluated through a comparative analysis of number of scan positions, point cloud completeness, and the time needed for manual scanning of outdoor construction sites. the proposed pipeline enables efficient collection of high-quality point clouds in outdoor environments, laying the groundwork for further artificial intelligence processing.