PSI - Issue 78

Simone Felicioni et al. / Procedia Structural Integrity 78 (2026) 1285–1292

1286

1. Introduction In recent years, the growing interest in robotics, computer vision, and immersive technologies has opened new frontiers in the field of structural health monitoring, enabling the development of immersive environments for the remote assessment of structural integrity. Traditional inspection practices are mainly based on visual inspections, requiring direct human access to structures and, thus, resulting to be time-consuming and potentially dangerous, especially in cases involving aging infrastructure or remote locations. These limitations highlight the need for developing more efficient and automated remote inspection methodologies. Virtual Reality (VR) is an emerging technology that enables users to interact with simulated environments in a fully immersive manner. By allowing operators to explore a virtual replica of a real-world structure from a first-person perspective, immersive VR enhances depth perception and situational awareness. This can be achieved by using head mounted devices (HMDs) that offer stereo viewing through small monitors placed in front of the human’s eyes. Over the past decade, VR systems have been largely adopted across several fields, such as education and training (Feng et al. (2020) and Lorusso et al. (2023)), or architecture and engineering (Sadhu et al. (2023) and Martins et al. (2024)). In the context of structural monitoring, VR allows experts to conduct visual assessments without the need for physical access to the site, thereby enhancing both safety and accessibility, particularly in dangerous or hard-to-reach areas, as in the end-to-end pipeline proposed by Pantoja-Rosero et al. (2023). Following this research direction, this paper proposes a pipeline for creating immersive environments to facilitate automated structural inspections by integrating data collected by robotic platforms and VR, with a focus on damage monitoring over time. It is particularly suited for use in heritage structures or civil infrastructure where regular monitoring is essential, but physical access is restricted or impractical. The first step in the proposed methodology involves the collection of a 3D map and its integration into a virtual environment, to enable users to remotely explore the site under investigation in an immersive setting. Periodic aerial flights are then required to acquire visual data to facilitate the identification and temporal tracking of structural anomalies such as cracks. To support the monitoring process, the framework introduces the concept of keypoint, or Point Of Interest (POI). In this context, a POI is defined as a spatial location of structural relevance, i.e. , a point in the 3D space associated with a structural defect such as a crack or deformation. Once identified, each frame containing structural damage is matched to a corresponding location in the pre-acquired point cloud using an image-based relocalization technique, and finally projected into the virtual environment. The keypoints are rendered in the virtual environment as interactive markers, that serve as spatial anchors for visualization: they can be selected by the user to show the sequence of images captured from the corresponding location, offering a time-aware perspective of damage progression. The proposed approach has been tested on a historic masonry church currently exhibiting significant cracking. The results demonstrate the system’s potential for supporting remote inspections, reducing reliance on manual surveys, and enhancing the structural behavior over time. 2. System Design Structural monitoring of buildings and infrastructure is a critical task in both civil engineering and cultural heritage preservation. The proposed framework aims to address the challenges of traditional inspection methods, that often rely on time-consuming, expensive, and potentially dangerous on-site visits. In this context, the integration of immersive visualization technologies with robotic platforms can significantly enhance inspection capabilities, allowing for safer and more frequent monitoring. In the presented methodology, a detailed 3D point cloud of the structure is first acquired using a Light Detection and Ranging (LiDAR) scanner. This high-resolution model is then imported into a Unity-based simulation environment, representing the basis of the virtual inspection space. Within this digital twin, users can freely navigate and explore the environment using a head-mounted display (HMD), simulating the experience of walking through the