Indoor Structural Defect Inspection Using Machine Learning-Enabled Augmented Reality Tools for Portable Devices
Part Of
IABSE Symposium Copenhagen 2026: Bridging Advanced Technologies - Structural Innovation
Journal Volume
2
Start Page
1220
End Page
1227
ISBN (of the container)
9798331335489
ISBN
9798331335489
Date Issued
2026
Author(s)
Togi, Ray Septian
Abstract
Visual inspections are essential for evaluating the integrity of building structures, yet traditional approaches often rely heavily on manual effort, resulting in inefficiencies, subjectivity, and potential inaccuracies. These challenges are compounded by the need for specialized personnel, which can introduce scheduling bottlenecks, increased costs, and delays that may affect safety outcomes. While sensing and automation technologies have advanced, their integration into real-time, spatially-aware inspection processes for indoor environments remains limited. This study presents a novel inspection framework that leverages augmented reality (AR) and machine learning to enhance the speed, accuracy, and usability of structural damage assessments indoors. By utilizing Apple’s ARKit and RoomPlan APIs, the system performs live spatial mapping and localization to generate detailed 3D reconstructions of interior spaces. Concurrently, the YOLOv8 object detection algorithm is employed to identify common structural defects such as cracks and spalling in real-time. These capabilities are combined within an interactive AR interface that allows users to record defect locations through raycasting and overlay them onto a floor plan extracted from the 3D model. Field evaluations conducted in rooms of varying sizes demonstrate significant improvements in accuracy and efficiency, reducing inspection times by 30–50% and lowering the manpower required. Additionally, the system automatically compiles images, 3D models, and annotated plans into a cohesive digital output, facilitating streamlined documentation and data management.
Event(s)
IABSE Symposium Copenhagen 2026: Bridging Advanced Technologies - Structural Innovation, 21 April 2026 - 24 April 2026, Copenhagen
Subjects
augmented reality
building inspection
defect detection
hands-on tool
machine learning
Publisher
International Association for Bridge and Structural Engineering (IABSE)
Type
conference paper