Automated defect inspection of concrete structures

Abstract

This paper presents a framework for automated defect inspection of the concrete structures, made up of data collection, defect detection, scene reconstruction, defect assessment and data integration stages. A mobile data collection system, comprising a 360° camera and a digital Light Detection and Ranging (LiDAR), is developed to render high flexibility of data acquisition of image and three-dimensional spatial data, while users traverse complex indoor environments. Deep learning algorithms are implemented to efficiently detect defects from the collected images, and a simultaneous localization and mapping algorithm is adopted for site reconstruction with the acquired LiDAR data. Based on the images of detected defects, assessment is conducted to evaluate the defect conditions, complemented with the defect dimensions estimated from the aligned image and LiDAR data. The position of defects could also be identified and mapped to respective structural elements. All the inspection results are finally integrated into existing Building Information Modelling files for better facility management. The proposed workflow was validated using a case study for determining concrete cracks and spalls in a real-world facility, successfully demonstrating the joint application of advanced technologies in facilitating inspection programs of civil infrastructure.