Jae-Joon Song

Accurately capturing internal crack evolution and deformation is essential for understanding rock failure and improving disaster prediction. However, traditional methods mainly focus on surface cracks and struggle to reveal the spatial characteristics of crack propagation, limiting insights into rock damage mechanisms. This study proposes a framework for automatically and visually characterizing crack behavior and deformation evolution by integrating in-situ X-ray computed tomography (XCT), deep learning, and digital volume correlation (DVC) techniques to address this limitation. This framework enhances crack identification accuracy and internal deformation field analysis. First, in-situ XCT scanning experiments were conducted to dynamically capture the crack propagation process in sandstone samples with an initial single defect. Next, a U-Net-based image segmentation approach was employed to improve crack extraction accuracy from CT images, enabling precise characterization of internal crack behavior. Finally, DVC technology was utilized to compute three-dimensional (3D) full-field displacement and strain distributions, facilitating the visualization and quantitative analysis of internal rock deformation mechanisms. The results indicate that the displacement difference between monitoring points on both sides of a crack enables the quantitative classification of crack types. Crack behavior in sandstone generally exhibits a mixed combination of opening, shearing, and tearing modes. Moreover, the heterogeneous displacement field distribution within the sandstone influences spatial variations in crack behavior. After crack formation, the internal strain field exhibits significant localization, which largely governs the sample’s failure process. The distribution characteristics of the ε _yy and ε _xy strain fields closely correspond to the internal crack morphology, demonstrating that their evolution effectively characterizes the internal rock damage process.

Accurate and efficient rock mass characterization is essential in geotechnical engineering, yet traditional tunnel face mapping remains time consuming, subjective, and potentially hazardous. Recent advances in digital technologies and AI offer automation opportunities, but many existing solutions are hindered by slow 3D scanning, computationally intensive processing, and limited integration flexibility. This paper presents Tunnel Rapid AI Classification (TRaiC), an open-source MATLAB-based platform for rapid and automated tunnel face mapping. TRaiC integrates single-shot 360° panoramic photography, AI-powered discontinuity detection, 3D textured digital twin generation, rock mass discontinuity characterization, and Retrieval-Augmented Generation with Large Language Models (RAG-LLM) for automated geological interpretation and standardized reporting. The modular eight-stage workflow includes simplified 3D modeling, trace segmentation, 3D joint network analysis, and rock mass classification using RMR, with outputs optimized for Geo-BIM integration. Initial evaluations indicate substantial reductions in processing time and expert assessment workload. Producing a lightweight yet high-fidelity digital twin, TRaiC enables computational efficiency, transparency, and reproducibility, serving as a foundation for future AI-assisted geotechnical engineering research. Its graphical user interface and well-structured open-source code make it accessible to users ranging from beginners to advanced researchers.

Accurate and efficient rock mass characterization is critical for geotechnical engineering. To achieve this, the digital rock mass characterization is rapidly growing. traditional tunnel face mapping faces significant challenges, including time-consuming manual processes, subjective geological interpretations, and safety risks for personnel in hazardous underground environments. Recent advances in digital technologies and AI offer opportunities for automation, but existing digital solutions suffer from time-consuming 3D scanning, computationally expensive post-processing, and limited flexibility of system integration. This paper introduces a face mapping solution named Tunnel Rapid AI Classification (TRaiC), which is an open-source MATLAB-based platform that integrates 360° panoramic (single-shot) photography, AI-powered discontinuity detection, 3D textured digital twin generation, rock mass discontinuity characterization, and Retrieval-Augmented Generation with Large Language Models (RAG-LLM) for automated geological interpretation and standardized reporting. TRaiC features a modular eight-stage workflow encompassing a simplified 3D geometric modeling, trace segmentation, 3D joint network analysis, and rock mass classification using RMR, with outputs structured for Geo-BIM integration. Initial performance evaluations demonstrate significant reductions in tunnel data processing and geotechnical expert assessments time. TRaiC produces a lightweight digital twin with a high textural fidelity but simplified geometry, ensuring computational efficiency and superior suitability for Geo-BIM integration. Its open-source nature brings about transparency and reproducibility of research and it constitutes the stepping stone for future research in the field of AI-assisted geotechnical engineering. Significant efforts have been made to name and organize scripts to bring maximum readability and flexibility to understand the functionality of functions, and a graphical user interface is designed to make it easy for the application and make it suitable for beginners, advanced, and experienced users to solve problems or develop it for research.

The structure from motion (SfM) and multiview stereo (MVS) techniques have proven effective in generating high-quality 3D point clouds, particularly when integrated with unmanned aerial vehicles (UAVs). However, the impact of image quality—a critical factor for SfM–MVS techniques—has received limited attention. This study proposes a method for optimizing camera settings and UAV flight methods to minimize point cloud errors under illumination and time constraints. The effectiveness of the optimized settings was validated by comparing point clouds generated under these conditions with those obtained using arbitrary settings. The evaluation involved measuring point-to-point error levels for an indoor target and analyzing the standard deviation of cloud-to-mesh (C2M) and multiscale model-to-model cloud comparison (M3C2) distances across six joint planes of a rock mass outcrop in Seoul, Republic of Korea. The results showed that optimal settings improved accuracy without requiring additional lighting or extended survey time. Furthermore, we assessed the performance of SfM–MVS under optimized settings in an underground tunnel in Yeoju-si, Republic of Korea, comparing the resulting 3D models with those generated using Light Detection and Ranging (LiDAR). Despite challenging lighting conditions and time constraints, the results suggest that SfM–MVS with optimized settings has the potential to produce 3D models with higher accuracy and resolution at a lower cost than LiDAR in such environments.