Natural disasters frequently inflict severe damage to the built environment, which demands a rapid, reliable, and cost-effective damage assessment for emergency response. However, traditional methods for post-disaster damage assessment often rely on static, labor-intensive data collection strategies that can be prohibitively expensive and struggle to adapt to dynamic post-disaster conditions. In this study, we propose a cost-aware Bayesian optimization framework combined with level-set estimation that continuously guides autonomous data collectors, e.g., an unmanned aerial vehicle (UAV), toward the most informative regions. By dynamically updating damage estimates across different geographic zones, our approach systematically reduces uncertainty while minimizing operational costs. The proposed framework is first validated using a controlled synthetic toy study, demonstrating the agent's ability to efficiently trace damage boundaries, recover the underlying damage map, and rapidly reduce predictive uncertainty. Furthermore, the approach is evaluated using high-fidelity disaster data generated by the Regional Resilience Determination (R2D) software. The results of the algorithm provide accurate and timely damage estimates that support informative and fast emergency response.
This article discusses the automated creation of a segmentation model from post-disaster aerial images for disaster analysis. Disaster damage assessment models must be efficient and accurate in order to provide relevant information on affected areas quickly. Manual design of such architectures is time-consuming and may not yield optimal results. To address this, we present a genetic algorithm that explores different components of UNet architectures to automatically design the best configuration. The goal is to maximize the accuracy of the segmentation model, which is the Mean Intersection over Union (MIoU) under explicit computational constraints. The algorithm explores a vast search space that contains various UNet architectural decisions (e.g., network depth, convolution type, oversampling strategies). It uses a fixed gene to generate a dynamic phenotype that serves as a neural network for the segmentation task. The results showed a significant reduction in model complexity, from over 40 million parameters for the State-Of-The-Art (SOTA) models to just 3.72 million parameters for our model. While maintaining good segmentation results, reaching 67.45% mIoU. We have successfully automated the design of the disaster damage assessment model by leveraging metaheuristic optimization and have provided a lightweight model that can be deployed in small devices for real-time disaster analysis.
Iyed Dhahri, Abdennour Azerine, Mahmoud Golabi et al.· International Conference on...· 0 citations
Wildfires generate rapidly evolving hazard landscapes that disrupt ground-based logistics and render conventional disaster-response operations ineffective, motivating the use of unmanned aerial vehicles (UAVs) in civil-protection missions such as medical resupply, casualty search-and-rescue, and perimeter surveillance. Existing evaluations, however, share a common limitation: they assess performance using navigation-centric metrics—primarily success rate—without accounting for the temporal value of the mission objective. This paper characterises, within each planner family, how a single risk coefficient ρ governs the trade-off between navigation success and time-decaying mission value: holding each family’s algorithm and replan trigger fixed and sweeping only ρ isolates its effect, so that the risk setting maximising a family’s navigation success need not maximise its mission value. To study this, FLARE is introduced, a deterministic benchmark for fire-landscape adaptive risk evaluation, inspired by recent Greek wildfire events; all hazard dynamics (fire spread, structural collapse, moving obstacles, dynamic no-fly zones) are modelled as benchmark abstractions rather than incident-specific reconstructions. FLARE evaluates eight planners spanning six risk-handling paradigm families, isolating the risk coefficient by sweeping ρ within each family (algorithm fixed) with A* as the risk-blind static reference, and quantifies mission impact—medication efficacy, casualty survival, and data freshness—through a strictly time-decreasing mission-score function. The mission-value leverage of ρ is strongly family-dependent: decisive for the incremental soft-cost-inflation family—whose success-optimal ρ collapses its mission value (0.91 → 0.41)—strong for the worst-case (CVaR) family, moderate for the sampling family, and negligible for the reactive, hard-threshold and frequent-replan families. In some families the success-optimal and mission-optimal ρ diverge—a within-family ranking inversion—while in others they coincide; because the algorithm is fixed across each sweep, the divergence is attributable to ρ alone. Success of navigation is therefore a necessary but insufficient proxy for mission effectiveness, and the risk coefficient is a meaningful per-family parameter for mission-aware configuration.
Konstantinos Zervakis, Ilias E. Panagiotopoulos· Automation· 0 citations
Environmental disasters, particularly earthquakes and floods, present immense challenges to humanity, demanding effective assessment and comprehensive management strategies. Given the increasing frequency and intensity of such events, their devastating impact on lives, the environment, and infrastructure calls for innovative approaches. Earthquakes can devastate entire urban areas within seconds, while flood events can also have long-term effects on the environment and livelihoods. Conventional disaster management methodologies often result in inefficient use of resources and suboptimal decision-making due to inadequate, near real-time, and cost-effective damage assessment capabilities. To address this critical gap, this paper introduces PostDisaster, an innovative platform designed to enhance disaster management and assessment and strengthen the resilience of the affected regions through the deployment of Virtual Reality (VR), Mixed Reality (MR), AI-supported satellite analysis, and drone-based tools. To overcome current technological limitations, the PostDisaster system provides decision-makers with tools for action planning and detailed damage analysis, delivering critical data at varying levels of detail. An AI-supported satellite image analysis tool delivers an immediate, macroscopic overview of the disaster’s scope. To obtain a more accurate overview of the situation, a drone-based tool generates a highly detailed 3D reconstruction of the affected area. which can be viewed in VR to enable decision-makers remote assessment of the situation and planning of measures. Finally, an on-site MR application allows users to walk through damaged infrastructure, mark damages and reconstruct the infrastructure. The reconstructed infrastructure with the marked damages can be subsequently viewed as digital representation in VR, ensuring comprehensive traceability of decisions for e.g., mission evaluations, debriefings or insurance assessments. This paper describes the current state of the PostDisaster system and the outcomes of the requirements analysis phase.
Elisabeth Broneder, Stefan Penn, Lukas Hofstetter et al.· AHFE International· 0 citations
Building safety inspections can be an impeding determinant of post‐earthquake recovery, yet their durations remain inadequately quantified because fine‐grained operational data are rarely available. Here, we present a probabilistic Bayesian framework to quantify building‐level inspection time, calibrated against a comprehensive multiagency dataset from the 2010–2011 Canterbury Earthquake Sequence (CES) in Christchurch, New Zealand. Using Markov chain Monte Carlo (MCMC) inference, we represent inspection timeframes as stochastic processes and infer marginal posterior distributions that jointly characterize aleatory variability and epistemic uncertainty in post‐disaster operations. The resulting posteriors distinguish the temporal signatures of rapid building assessment (RBA) and detailed damage evaluation (DDE) protocols and reveal strongly upper‐tailed delays driven predominantly by institutional and logistical frictions beyond damage state alone. By providing empirically derived probability curves for inspection time, the framework enhances the fidelity of regional recovery simulations and offers emergency managers an evidential basis for capacity planning, resource mobilization, and inspection strategies designed to accelerate community recovery after future earthquakes.
Lianyan Li, A. Chang-Richards, Megan Boston et al.· Earthquake spectra· 0 citations
This work presents HASTE (High-speed Assessment and Satellite Tracking for Emergencies), a no-code web platform that lets analysts who are not machine learning engineers produce per-building damage maps from post-disaster satellite imagery, and describes the platform, methods, and the engineering that supports them.
Caleb Robinson, Anthony Ortiz, S. Nsutezo et al.· 0 citations
Evacuation planning for disaster preparedness requires making critical decisions under uncertainty before the number and spatial distribution of evacuees are known, including shelter location, evacuation route assignment, and relief supply prepositioning. Because these decisions are highly interdependent, planners must balance the competing objectives of maximizing relief demand coverage and minimizing evacuation time. We propose, to our knowledge, the first adaptive robust evacuation planning model to jointly optimize shelter locations, evacuation route assignments, relief supply prepositioning, and post-disaster relief item distribution. The model minimizes the worst-case weighted sum of unmet demand for relief items across shelters and the congestion-dependent evacuation time. We characterize theoretical complexity drivers of the resulting problem with mixed-integer recourse and develop a partition-and-bound algorithm that maintains tractability by selectively partitioning only the most critical subpartition of the uncertainty set while producing strong upper and lower bounds. To quantify the value of centralized route planning, we also formulate a user route choice alternative in which evacuees choose among acceptable routes. Computational experiments quantify the value of centralized route planning, which reduces worst-case unmet demand and evacuation time by up to 90.6\% and 79.3\%, respectively, relative to decentralized user route choice. Adaptive post-disaster supply redistribution further improves relief demand coverage. Coordination between evacuation routing and relief distribution creates substantial operational value under uncertainty. Centralized route planning primarily mitigates congestion by coordinating evacuee flows across shelters, whereas adaptive redistribution primarily improves relief demand coverage when relief supplies are scarce or inflexibly prepositioned.
Jaehyuk Kim, Merve Bodur, Maria E. Mayorga et al.· 0 citations