Large-scale language models (LLMs) have spread rapidly since the release of ChatGPT, and many research reports have been published on the ability to understand and predict the mental state of others, such as their intentions, beliefs, desires, and knowledge, as represented by ToM (Theory of Mind). While the ToM field is expected to be applied to civil engineering fields such as urban planning that takes into account people’s personality, behavior, and feelings, there is a problem in that it is difficult to collect data on behavior and feelings and personality data at the same time. In this study, we prepared a dataset by simultaneously conducting a questionnaire and personality diagnosis, and verified whether LLMs can estimate the personality of the questionnaire respondents. Although the accumulated data is still not large, at around 200 people, it was confirmed that the personality traits of the respondents can be estimated with a certain degree of accuracy even from relatively short sentences such as questionnaire surveys, and it became clear that there is a high possibility of using LLMs in fields such as the analysis of human behavior and psychology.

Promotion of AI using in the civil engineering field strongly requires digitization and digitalization. Data is not only made by sensor but also by human imputing. It is not realistic to detect data with different results before and after processing due to human intervention by multiple human checks. This study evaluated the feasibility of detecting such inconsistent data under realistic conditions using ChatGPT, an AI chatbot with a large-scale language model. Experiment used local government data from xROAD’s road bridge data for the experiment. Discrepancies between data obtained from the public API including multiple bridges and inspection reports were detected in a short period of time and with almost fixed prompts.

The aging of bridges, coupled with the shrinking of public financial resources due to population decline, has increased the maintenance and management burden on municipalities. On the other hand, with the expected decrease in traffic demand for certain bridges as a result of population decline, it has become necessary to formulate bridge abolition plans that are tailored to the characteristics of each municipalities. Therefore, this study conducted a case study on 81 municipalities across four prefectures in the Hokuriku region, attempting to categorize municipalities based on their bridge maintenance status to support the formulation of abolition plans using quantitative indicators and to establish cooperative frameworks among similar municipalities. Furthermore, this study proposes the concept of the “Bridge Abolition Effectiveness” as a quantitative indicator to demonstrate the effectiveness of bridge abolition. Through analysis using open data, distinct characteristics of the bridge maintenance conditions were revealed for the categorized groups of municipalities. Additionally, by applying the Bridge Abolition Effectiveness, it was possible to identify municipalities that may be suitable candidates for considering bridge abolition.

In recent years, there has been growing interest in using digital twins constructed from point cloud data measured in real-world environments. In Japan, efforts are underway to create digital twins of the national territory using measurement data such as point clouds. However, continuously updating the entire shape of the national territory using all measurement data incurs substantial costs. Therefore, it is necessary to identify techniques that can efficiently update only the areas where changes have occurred, detected from point cloud data of different periods. A recently proposed method detects differences by voxelizing the entire space surrounding the point cloud data and using a Merkle tree. However, this approach cannot detect changes at arbitrary resolutions. To address this problem, our study introduces a method that generates hash values for each hierarchy of the Merkle tree, enabling the detection of changes in point cloud data between two periods at arbitrary resolutions.

abstract

The declining birthrate, aging population, and increasing motorization are making it difficult to maintain public transportation systems due to a decline in supply capacity and a decrease in the number of users. Therefore, there is a need for an efficient public transportation network that a can improve transportation performance with less supply capacity, based on an understanding of the demand for bus use in relation to the people flows. To evaluate the consistency between the current bus routes and frequencies and the people flows, we analyzed the relationship between the data of transportation performance by bus route system and the amount of people moving by using people flows data. Based on the results, we discussed the factors that contribute to the high or low efficiency of bus routes in terms of routes, frequency, and land use. The results also suggest the need to establish traffic nodes and bus lanes, and to reduce the number of competing routes in order to improve public transportation efficiency.

Although resident evacuation behavior has been analyzed from various perspectives, the number of casualties due to heavy rain disasters continues to increase. Therefore, it is still difficult to say that the issues related to resident evacuation have been resolved. As a new attempt, the authors attempted to use XAI (Explainable AI) to analyze the factors behind resident evacuation behavior, but there are still issues with the predictive accuracy of the behavior model.

In this study, questionnaire survey data on resident evacuation behavior during heavy rain disasters was converted into image data and XAI (Explainable AI) was applied to image recognition technology. As a result, the predictive accuracy of the resident evacuation behavior model was improved. In addition to having experience of disaster before a disaster, we showed that obtaining appropriate evacuation information during a disaster and damage to one’s home affect evacuation behavior.

Ecotourism has problems such as accidents caused by participants participating in tours that are not suitable for their physical fitness, and overtourism. Digital archiving of eco-tour route trails is considered useful for accident prevention by visualization of information on difficult spots along the trails, as well as for quantitative and continuous monitoring. In this study, we measured the trail to Kura Falls on Iriomote Island using LiDAR, and selected difficult spots based on the minimum height of each 5m section and the calculated difference of the 85th percentile value from the minimum. In addition, we calculated and visualized the difference between the two periods, and revealed changes in fallen trees, etc. of about 0.15m. The application of LiDAR to the digital archive of mountain trails showed that it is possible to select difficult areas and to conduct primary screening for natural environment monitoring surveys.

This paper proposes an alternative model for inundation area prediction using conventional numerical methods by applying dimensionality reduction techniques to inundation area prediction using deep learning. For dimensionality reduction methods、 used singular value decomposition (SVD), non-negative matrix factorization (NMF), and auto encoder (AE) are employed. The proposed method was applied to a simulation of Arakawa River inundation prediction, and its validity and effectiveness were examined from the viewpoints of computational time and accuracy. As a result, it was confirmed that the proposed method can significantly reduce computation time and maintain accuracy. We also confirmed that SVD is the most effective of the three dimensional compression methods employed.

Fatigue cracks in bridges are becoming more and more apparent every year and are a problem for many bridges. Fatigue cracks are a failure phenomenon that progresses gradually due to repeated stress and deformation and can have a significant impact on the safety of the structure. Therefore, it is important to detect fatigue cracks at an early stage and to perform appropriate repair and reinforcement. In this paper, experimental data from vibration fatigue tests conducted in the past is reused to investigate the possibility of simplifying the process of crack detection by automating stress calculation and graphing for the numerical processing of a large amount of data. We also plan to compile a database of this data and develop it into a machine-learning system for predicting crack propagation by studying trends in stress states during crack initiation and propagation.

Currently, the number of aging bridges is rapidly increasing in Japan, and there is a need to establish a more efficient and quantitative bridge inspection method. The authors have focused on the impact vibration testing using a portable FWD and have conducted experiments on small bridges. We have measured the response acceleration and applied the MTS method to the difference in response acceleration between multiple measurement points to determine the localized damage and deterioration in the slab. In this study, we used FEM analysis to investigate the influence of the degree and extent of deterioration on the results obtained using the MTS method, and the usefulness of the MTS method for detecting deterioration and damage in slabs.

In urban rivers where combined sewer overflow inflow occurs, scum forms after rainfall, causing prob- lems such as bad smells, deterioration of the landscape, and adverse effects on the ecosystem. As a coun- termeasure, early warning through monitoring of scum is desired. In this study, we examined the monitoring and image recognition requirements with the aim of detecting scum formation at an early stage, and then used data from a river monitoring camera on the Hirano River in Osaka Prefecture to experimentally clarify the image recognition model and loss function that match the characteristics of scum images. Based on the experimental results, we added images with low accuracy in detecting scum to the training data and con- ducted final training, and developed a high-precision scum detection model that satisfied the requirements by achieving Scum IoU95%.

The Ministry of Land, Infrastructure, Transport and Tourism (MLIT) has set a new urban policy to create people-centered spaces in the city where people feel comfortable and want to walk around. In order to achieve this, it is necessary to evaluate the comfortableness of public spaces, but this task is carried out by manual survey, which is problematic because it requires an enormous amount of labor.

In this study, we proposed an automated method for surveying the comfort of public spaces using a multimodal model. To confirm the validity of the proposed method, we compared the accuracy and effort of the survey with a visual survey. Based on the validation results, we discuss the applicability of the pro- posed method to actual work and future issues.

This study applied TimesNet (one of the neural network models) to long-term, large-sample-size, and time-series data of water levels, observed in drainage management with pump operations in a low-lying watershed. Time series data, in general, contain complicated and multiple patterns. TimesNet is a prediction method that can automatically extract multiple periodic features, contained in the input data, using spectral analysis, then transforming to two-dimensional information that arranges continuous data according to these periodic features, and finally predicting future continuous data. Applied to long-term water level data, this method performed worse in prediction accuracy, when compared with conventional prediction methods. However, it was confirmed that the method better reproduced flood waveforms during the specific drainage period involving the largest flood event.

In the developing country of the Democratic Republic of Timor-Leste, heavy rains during the rainy sea- son cause frequent flooding. Cyclone Seroja in 2021 caused extensive damage to roads, bridges, and other infrastructure. There have been active efforts to strengthen disaster prevention infrastructure. However, as a developing country, Timor-Leste lacks data on disasters, and no assumptions have been made regarding the damage caused by floods. Therefore, in this study, urban development and flood risk assessment using GIS and remote sensing were conducted in Cristo Rei Hera, Dili, the capital city of Timor-Leste. Using Sentinel-2, a satellite operated by the European Space Agency, and Arc GIS Pro, a GIS software, a land use and land cover classification was conducted to compare the area of urban areas overlapping inundated areas. As a result, the urban area of the Hera district increased by about 4 km² between 2016 and 2024, and the inundation area approximately doubled. It was found that the urban area in the Hera district is spread out in areas of high flood risk.

In recent years, the number of vacant houses in Japan has been increasing rapidly due to the declining population, and the government has been proactively developing measures to solve the problem of vacant houses and providing subsidies to encourage the expansion of vacant house utilization projects. On the other hand, when utilizing vacant houses, it is necessary to focus on the regional characteristics of each vacant house and the owner's intention. In this study, we use data from a survey of vacant houses conducted by Hatoyama-Town, Hiki-Gun Saitama Prefecture, to quantitatively evaluate variables that affect the ease of utilization of vacant houses, taking into account the characteristics of the surrounding area, the condition of vacant houses, and the intentions of their owners. Through the analysis in this study, it was realized to develop an index that can be used as a standard when a municipality utilizes vacant houses.

If a large-scale disaster occurs and there is a shortage of pharmaceuticals, the inability to pre- scribe necessary medications to patients with chronic diseases increases the likelihood of their con- dition worsening. This study aims to estimate the future regional demand for pharmaceuticals in preparation for a sudden increase in demand due to a large-scale earthquake disaster. Focusing on Hakui City in Ishikawa Prefecture, we first examined the pharmaceutical prescription trends over the past ten years and then projected future prescription trends. This study specifically targets anti- hypertensive drugs, antiepileptic drugs, and antidiabetic drugs, which are critically necessary for patients with chronic diseases. Using the SARIMA model, which is employed for time series data estimation, we projected the number of prescriptions under normal circumstances for the next five years. The analysis results showed that the number of prescriptions for antihypertensive drugs and antiepileptic drugs decreased by approximately 34.2% and 42.1% respectively over the five years from the start of the estimation period, while the number of prescriptions for antidiabetic drugs de- creased by only about 4.2%.

In Wajima City, Ishikawa Prefecture, which was hit by the 2024 Noto Peninsula Earthquake, many residents have been forced to live as evacuees for a long period of time due to the pro- longed restoration of infrastructure. In this study, time-series clustering of the nighttime pop- ulation trends in 26 areas in Wajima City obtained from the KDDI Location Analyzer was conducted, and each area was classified based on the medium- to long-term trend of the nighttime population flow. As a result, we succeeded in visualizing the time lag between the time of the earthquake and the peak of the nighttime population, and the progress of recovery and reconstruction plans. The difference in the speed of recovery of the nighttime population between regions indicates the time required for the return of the resident population and pro- vides insight into the current status of the affected areas during the recovery process.

The Hokuriku Shinkansen line extension between Kanazawa and Tsuruga opened on Saturday, March 16, 2024. The time-saving effect of the Tsuruga extension is expected to significantly increase the number of people in the Hokuriku region who travel to and from other regions. However, there is a concern that the opening of Tsuruga Station will reduce the convenience of travel to and from the Kansai and Chukyo regions, as passengers will have to change trains at Tsuruga Station. In this study, we conducted a questionnaire survey of Kanazawa citizens in Ishikawa Prefecture prior to the opening of Tsuruga Station and analyzed the effects on users of the Hokuriku Shinkansen Line of the change in transfer times and fares resulting from the opening of Tsuruga Station. The study revealed that changes in transfer time are affected by age, annual income, frequency of travel to the Kansai region, and resistance to travel to the Kansai region. It was also found that the change in fare was affected by the frequency of travel to the Kansai region and the sense of resistance to travel to the Kansai region.

In Japan, the acceptance of international cruises is resumed from 2022, and the number of port calls is expected to gradually recover to the level before COVID-19. In addition, the Ministry of Land, Infrastruc- ture, The New Tourism Nation Promotion Basic Plan has formulated a plan to revive Japan's cruise industry and has set goals for a recovery in the number of cruise visits to Japan. Cruise tourism is expected to become an important tourism industry in Japan. In cruise tourism, it is considered that the amount of money spent by passengers on tourism is a major factor in the economic impact of the port of call. On the other hand, the amount of money spent by passengers at the Port of Yokohama, which is often used as a departure and arrival port, tends to be lower than at other ports of call, despite the fact that passengers stop there twice for embarkation and disembarkation. Therefore, in this study, the tourist behaviour of cruise passengers at ports of arrival and departure is analysed by questionnaire surveys of passengers, and we obtain knowledge on how to make the cities that are also ports of arrival and departure more attractive to passengers for sight- seeing.

This study examines the statistical estimation methods for multivariate analysis parameters in Effective Stress Dynamic Analysis (LIQCA). The constitutive model used in LIQCA include several fitting parameters that cannot be directly observed in the field. he accuracy of setting these fitting parameters often depends on the analyst’s experience. Therefore, to objectively describe the reliability of the analysis, it is necessary to clarify the process of setting these parameters. In this study, we collected and organized the parameters and soil test results used in the parameter-setting process from 27 LIQCA analysis cases conducted by experts in this field, and constructed a database. Using the statistical information of parameter sets obtained from the database as prior information, simple Bayesian linear regression was applied as a reference solution. Furthermore, clustering was performed on the database, and linear regression using hierarchical Bayesian methods and sampling using similarity evaluation were applied to discuss the impact of the estimation results on the uncertainty of the numerical analysis results. Based on these considerations, this paper proposes an effective framework for utilizing the experience of analysts in setting material parameters and discusses the problems in promoting these database approaches.

In recent years, the global race to develop quantum computers has intensified. Quantum computers are expected to be applied across various engineering fields, with particularly high expectations for their integration with High Performance Computing (HPC) in the field of the computational mechanics. However, the development of computational mechanics techniques utilizing quantum computers is still unexplored. Therefore, in this study, we conduct a fundamental investigation of the Boundary Element Method (BEM) using quantum algorithms. First, we briefly describe the current state of quantum computing. Then, the general formulation of BEM for 2D Laplace equation targeted in this research is discussed. Subsequently, from an engineering perspective, the quantum algorithms required for solving the BEM for 2-D Laplace equation using a quantum computer are explained. Finally, by presenting numerical examples obtained by using a quantum simulator, we demonstrate the potential for implementing the BEM on quantum computers.

In post-earthquake inspections of road bridges, it is necessary to quickly and accurately assess the damage to decide on a course of action. The current post-earthquake inspections are almost always conducted by visual inspection, and there are no objective standards for determining necessity of additional investigations. However, there is a possibility that damage that cannot be recognized by visual inspection may influence the decision on the course of action. In order to improve the certainty of the course of action, it is necessary to clarify the information on damage to be collected, and to provide a means of acquiring information that contributes to this information and an objective evaluation method for it. As a case study, this study conducted a visual inspection of the superstructure of a steel arch bridge damaged in the Noto Peninsula earthquake, and also conducted point cloud measurements. The deformation of the members was measured using a model created from the point cloud in order to detect the distortion of the superstructure, and the results were quantitatively evaluated.

Heavy rainfall events frequently cause flooding, leading to road blockages in affected areas. These blockages hinder evacuation and recovery efforts, making it crucial to rapidly assess road conditions of the roads. In this study, we detect flooded areas using aerial imagery and estimate road flooding by overlaying the detected flooded regions with road data. Despite the challenge of false positives in areas with significant sediment and river regions, the detection of flooded road sections was generally successful. By overlaying the detected flooded regions with road data, it became possible to visualize the road flooding locations. The detection showed an error range of 0.4 to 8.4% in regions with low actual flood rates, indicating its utility in determining road passability.

This paper describes the results of efforts to resolve issues that have previously been associated with measurements using multicopter drones, such as reduced work efficiency when measuring wide areas due to short flight times, or difficulty in measuring distant objects. Specifically, the paper describes the results of developing a drone that flies while being powered by a gasoline-powered generator, and an algorithm for analyzing the big data acquired. This paper shows that green laser measurements using the developed long-flying drone can obtain highly accurate elevation values, and also demonstrates the usefulness of this algorithm in quantitatively visualizing changes in topography from point cloud data obtained at two different times in a case where 3D measurements of the seafloor topography of a coastal area were conducted.

In this study, we propose a method for the high-precision automatic generation of findings for distress images. Recently, multi-modal models have attracted attention as generative AI since they are capable of understanding both images and texts with high accuracy. In addition, they can learn and adapt to various tasks with only a few input examples. Therefore, in this paper, we propose a method to efficiently learn the relationship between distress images and findings based on the multi-modal model. We obtain these pairs based on similar image retrieval. This approach enables highly accurate finding generation. We also use the structural components and types of damage that engineers refer to when creating findings. By compressing the data pool for retrieval using this information, we can acquire more useful pairs of distress images and findings. In the last of this paper, we confirm the effectiveness of the proposed method through experiments by generating findings for distress images contained in actual bridge inspection reports.

The goal of this research is to establish a topology optimization method using quantum annealing that can search for a better optimal solution at high speed without falling into a local optimum. As a fundamental step, we have implemented and verified a method for topology optimization of two-dimensional trusses using quantum annealing and generalized Benders decomposition. In particular, the method was verified under conditions that have not been used in previous studies, and the effects of various calculation conditions on the results were evaluated. The verification results show that the proposed method is capable of performing reasonable optimal calculations, and that the penalty coefficient for the volume constraint and the setting of the volume change in the iterative step of the optimal calculation process are particularly important to obtain reasonable calculation results with the proposed method.

This study compares bridge deformation analysis based on time-series InSAR with structural simulation based on finite element analysis for a hinged PC bridge, and discusses the accuracy verification of InSAR displacements. In the past, the steel bars supporting the girder ends of this PC bridge broke, resulting in steps. To efficiently extract the structure-derived InSAR displacements from the limited number of reflection points on the bridge, a temporal displacement modelling method using temperature variation is employed. As a result, the following was found: 1) The SAR vertical displacement per 1 °C on the central span up to three years before the step event was close to the vertical displacement per 1 °C in the structural simulation, 2) In the time-series displacement of the central hinge, a deviation from the simulation results was observed two years before the event, with a gradually increasing trend of sagging.

While Kanazawa City is aiming to realize a compact city for sustainable urban growth, the city’s past urban development has been overly dependent on automobiles, which has caused the city’s liveliness to flow out to the suburbs and the vitality of the central city area to decline.

On the other hand, Kanazawa University’s Kakuma Campus, the subject of this study, is located in the suburbs far from the center of Kanazawa City, and it is not easy to get around the campus or to the center of town.

In order to gain knowledge to change students’ mobility in a socially and personally desirable direction, we conducted an experiment and analysis to improve the convenience of student mobility by expanding the area of the Kanazawa City public shared cycle “Machi-nori”. As a result, we found the possibility of changing students’ behavior (breaking away from dependence on automobiles, and promoting their mobility in the city center).

Water facilities such as reservoirs are essential to our daily lives. The entire reservoir may be compromised if a reservoir’s embankment partially collapses or leaks. However, maintaining, managing, and inspecting reservoirs is a major social issue.This study focuses on the partial collapse of embankments to improve the efficiency of reservoir inspections. The purpose of this study is to develop an automatic detection method for anomalies, specifically targeting the partial collapse of reservoir embankments using aerial images. We propose a method to automatically detect partial embankment collapses employing an autoencoder for unsupervised learning.Additionally, we propose a method to quantitatively indicate the location and size of abnormalities. The effectiveness of the proposed method has been demonstrated through actual experiments using a UAV in a real reservoir environment.

In this study, the method of noise removal in 3D point clouds using sensor fusion of LiDAR and camera was introduced. First, a point cloud measurement was performed using Matterport Pro3 in Fukushima Prefecture, Japan. The bridges and other man-made objects were scanned while people were also scanned. The relative positions of the camera and LiDAR were adjusted using a checkerboard for cross-calibration. Internal and external parameters were then obtained to map the 2D image to the 3D point cloud. This allowed us to correlate each point in the point cloud with a pixel in the image and develop a segmentation method for noise removal. Next, we extracted the region of a person in a 2D image and mapped it to a 3D space to classify noise points related to the person in the point cloud. It was confirmed that this method can effectively classify people even in the presence of many other objects in the point clouds. The evaluation results using recall, precision, and F1 scores had mean values of 0.923, 0.878, and 0.889 for all samples, respectively, indicating that highly accurate noise reduction is possible. The results of this study are expected to be effective for cleaning and pre-processing 3D point clouds. Future perspectives include evaluating the accuracy of this method on data with a larger number of people.

Maintenance work on railways, such as track maintenance and inspection and the maintenance of ledgers, is generally carried out by engineers on site, placing a heavy burden on the maintenance site. Against this background, systems have been developed to determine the degree of deterioration of wooden sleepers. In this study, we developed a track component identification model to identify various types of sleepers and rail fastenings, with the aim of centrally assessing the condition of track components. In addition, we developed kilometreage correction method to improve the accuracy of kilometreage estimation by utilising depots used to correct the position of track inspection car data. Furthermore, we developed ‘track components condition evaluation system’ by introducing these methods into the previously developed wooden sleeper deterioration evaluation system and proposed an efficient track member management method utilising this system.

In this research, a damage detection method for the ceilings of buildings is proposed, utilizing Anomaly Detection by Efficient GAN, which enables anomaly detection based on images taken under normal conditions. The effectiveness of this method has been verified through empirical experiments conducted on the ceilings of actual buildings. The proposed method determines the presence of damage when the Anomaly Score exceeds a threshold established based on the Anomaly Score from normal conditions. Additionally, as the method of obtaining images of the target areas of buildings significantly influences the convenience of the damage detection method, both the use of UAVs and UGVs have been considered. The results of the investigation confirm that using UGVs is more convenient indoors due to the ease of autonomous navigation. This research demonstrates that UGVs offer higher practicality for indoor applications, thereby enhancing the efficiency of the damage detection process. The findings suggest that the proposed method, combined with the appropriate image acquisition means, provides a robust solution for the automated detection of ceiling damage in buildings.

The accuracy rate of sediment disaster warning information in Japan is extremely low, around 5%. The first reason for this is that data consisting of hourly rainfall and soil water index are processed as independently generated data points, not as time series. The second reason is that the RBFN output values, which are threshold values calculated based on the independent data, are not linked to the occurrence of sediment disasters. Therefore, in order to improve the accuracy of this warning information, it is necessary to construct a statistical model with time series data of rainfall from the beginning of rainfall to the occurrence of a disaster, and then evaluate whether the output is associated with the occurrence of a disaster. This study presents a method for detecting the singular values included in the rainfall that causes sediment disasters using the general state space model as one of the proposals, and verifies the validity of the method.

This study examined the applicability of a quantum-inspired black-box optimization for determining post-disaster bridge restoration planning in road-bridge networks. The total travel time, obtained using the user equilibrium assignment method, was defined as the function of the network. The objective of this optimization problem is to minimize the area of the resilience triangle. A black-box optimization combining Fujitsu’s digital annealer, genetic algorithms, and factorization machines based on machine learning was employed for the optimization method. The optimal restoration plan was explored using this combined method and a standalone genetic algorithm, commonly applied in past research, for a small-scale virtual road-bridge network including 14 bridges. As a result, the combined method demonstrated an advantage in terms of search efficiency and analysis time, with faster convergence in the early stages of the search compared to the standalone genetic algorithm.

Recently, river bridge piers have been damaged by score due to flooding. Considering recent damage cases, we are developing an extraction method adopting machine learning model to extract bridge piers which may be damaged by scour. In this study, we examined the applicability of machine learning model trained on dataset collected by RTRI to recent damage cases. In parallel, we analyzed various parameters included in missed cases by machine learning model. As a result, machine learning model can extract recent damage cases which are not included in training data. And we show the characteristics of the piers included in missed cases by machine learning model.

To ensure safe and smooth road usage, it is necessary to appropriately detect abnormalities in road auxiliary structures, including road information boards. In this study, we conducted a year-long acceleration measurement on the road information board supported by type F pillars and analyzed the characteristics of the eigenfrequency of these boards over a long period. The results showed that even under normal conditions, the natural frequencies tend to change due to variations in the temperature of the steel material, which affects the performance of anomaly detection. To suppress the changes in natural frequencies caused by temperature variations, a correction method using simple linear regression was implemented, resulting in noticeable improvements. Next, we applied the Mann-Whitney’s U test to the time series of natural frequencies to detect their temporal changes. Unfortunately, many false positives were estimated because of the fluctuations of the natural frequencies. So, we estimated the cumulative distribution function of the Z-value in health condition and detected the abnormalities by evaluating the occurrence probability of the Z-value with the estimated function. The results showed that the change in natural frequency of 0.01 Hz was detected with 95 % accuracy.

This paper proposes searching for missing persons in large-scale natural disasters using ground-penetrating radar and airborne SAR. First, we experimentally confirm the detection of ground objects on a sandy beach using 800 MHz ground-penetrating radar and theoretically calculate the size and depth of the objects that can be detected using the FDTD method. Next, experiments on detecting ground objects by L-band aircraft-mounted SAR are described. Finally, we describe the number and type of objects detected on the ground and the depth of detection from September 2014 to March 2020 by applying the FDTD method to the actual search for missing persons on the Yuriage coast of Natori City, Miyagi Prefecture, where many people went missing after the tsunami disaster caused by the Great East Japan Earthquake.

This paper describes the automatic detection of beach debris using an unmanned aerial vehicle (UAV) and deep learning to efficiently search for missing persons in large-scale disasters such as tsunamis. Using YOLO (You Only Look Once), capable of real-time detection as deep learning, we examine the detection rate of beach debris based on three years of long-term observations at a sandy beach. In addition, we describe a method that introduces background classification as a pre-processing step to apply the method to various beaches, including sand, gravel, and vegetation. The results show that an average detection rate of 85.4% can be obtained for three years between 2020 and 2022 by learning only from UAV aerial images taken two months in 2020 and that background classification can improve the detection rate by 17.2%.

Although traffic detector data is often used to understand road traffic state, traffic detector is mainly installed at main traffic signalized intersection. Therefore, it is known that understanding the traffic state has difficulty at the area which traffic detector is not installed with high density. This paper describes the possibility to understand road network traffic state using bus prove data which is operating with same time table and route even if it is rural city. Especially, this paper is focusing on the comparison of the characteristics of Macroscopic Fundamental Diagram drawing by traffic detector data and bus prove data as case study in Kochi city. In addition to that LSTM model which is known as the one model of machine learning approach has been applied using collected bus prove data. The results of developed LSTM model showed that bus prove data usability of road network traffic state analysis since it can understand the change of traffic state at congestion period on weekdays.

To study the technology for detecting serious damage to structures by satellite, 275 bridges were selected and analyzed from bridges over 50 meters in the Noto Peninsula and Kanazawa Uchinada area where the 2024 Noto Peninsula Earthquake occurred. As a result, anomalies in many of the bridges have been detected where damage occurred. On the other hand, in some cases, the damaged areas were not visible to the satellite due to typical reasons such as the influence of mountains, trees, and vegetation. While the results of this analysis do not answer the need for rapid information provision immediately after an earthquake, they do demonstrate the potential of satellites for wide-area analysis. The ability to detect anomalies with less post-earthquake observation data is a major challenge.

Road Eye is a special vehicle for inspecting various road conditions under driving and it is mainly employed in highway at the east area in Japan. One of the measured quantities is a continuous road surface image captured by the line cameras and health of the road surface is manually evaluated from state of cracks in the image according to several complex rules. The final objective of the study is to automatically evaluate health of the road surface by applying several deep convolutional neural networks to the multi-scale images, and as the first step of the study, this paper presents two neural networks for semantic segmentation. The first one is a network for distinguishing road surface, expansion-joint and grooving in the image in wide area. Then, from the road surface in local area, the second one detects and classifies cracks according to its depth. Both networks show accuracies equivalent to the manual judgements.