Automated Bridge Inspection Using Mobile Ground Robotics

摘要

The use of mobile ground and aerial robotics presents a powerful means to augment current visual inspection practice by overcoming some common weaknesses related to accessibility, repeatability, hidden defect detection, and quantification. In this paper, a novel ground robotic bridge inspection platform consisting of a rugged mobile platform equipped with an onboard computer and several calibrated and time-synchronized sensors is introduced. This platform, along with custom localization and mapping software, is shown to produce high-quality 3D point cloud maps for the underside of typical concrete bridges, which are often the most challenging to inspect. The quality of these maps is compared to ground truth using terrestrial laser scanner measurements and the maps are shown to have an overall scale error of just 1.3%. The maps from the proposed system can be generated in real time, while continuously scanning the bridge, resulting in a significant reduction in the inspection time compared to using a terrestrial laser scanner (TLS). Additionally, this work presents a novel procedure for fully automated point cloud colorization and semiautomated defect localization and quantification. The former allows for realistic visual renderings of the bridge for remote inspection by offsite inspectors. The latter allows for increased accuracy of defect quantification compared to traditional inspections, while eliminating subjectivity between inspections and inspectors.