Jul 2026· Signal Processing and Communications Applications Conference· pp. 1-4· 0 citations· 10 references
Abstract
This paper presents a modular decision support system that infers the primary location of the user or the reported incident and a situation-aware risk level from multi-turn Turkish disaster dialogues between a help-seeking user and an AI-supported emergency assistant. The assistant guides the user with follow-up questions about health status, number of affected people, structural damage, environmental hazards, and known nearby landmarks to complete missing information. The system manages the dialogue with a finite state machine, determines the location by linking user cues to a local GeoJSON point-of-interest database and by landmark verification, and produces explainable risk scores with Multi-Criteria Decision Analysis. The key novelty is treating landmarks as an evidence layer that verifies the current location hypothesis through proximity and clustering instead of directly replacing candidates based on a landmark signal. In a ten-scenario pilot evaluation, accuracy, end-to-end latency, and token usage are reported for three configurations.
These constitute the most salient issues in disaster preparedness which are rapid evaluation of the dangers and judicious application of precautionary warnings. Traditional models, that utilise the physical implementation of the sensor webs and that typically use inflexible threshold logic, typically lack the structuring to query dynamic situational contexts or give results in a subtle way of interpretation. This paper will present a softwarefocused model of disaster management wherein me-teorological, media, crowdsourced social, and archival coverage of the disaster are combined into an integrated analysis channel. The scoring scheme is deterministic and produces a Composite Risk Index by the algorithmic combination of all streams of data, a big language model then provides post-hoc, natural language explanations in order to justify danger ratings. The modern risk environments are modelled in the interactive dashboard which is designed based on latest web-based technologies and sends directive advisories. Simulations of multi-hazard scenarios with the use of empirical trials identify a classification accuracy of 91.4 percent and end-to-end processing time that is an average of twelve seconds, thus proving the applicability of the system in operational conditions limited by small resources, damaged infrastructure, or inaccurate sensors.
Rudrala Jahnavi, K. S. Upendra, M. Mujeebuddin et al.· 2026 5th OPJU International...· 0 citations
Natural disasters continue to create significant challenges for emergency response teams due to delays in incident assessment, inefficient allocation of rescue resources, and lack of coordinated decision making. Conventional disaster management systems mainly depend on manual operations, making it difficult to respond effectively when multiple emergencies occur simultaneously. This paper presents a Multi-Agent Disaster Management Simulator that automates the disaster response process using intelligent software agents, machine learning, graph-based routing, and generative artificial intelligence. The proposed system consists of four autonomous agents responsible for incident detection, resource allocation, rescue coordination, and performance evaluation. A Random Forest model predicts disaster severity using environmental and demographic features, while the A* search algorithm identifies the shortest rescue path between response teams and disaster locations. Google Gemini API is integrated to generate human-readable explanations for resource allocation decisions. A React-based dashboard provides live visualization of incidents, rescue routes, and performance metrics through WebSocket communication. Experimental results demonstrate that the proposed approach improves response efficiency, reduces manual intervention, and provides better transparency during emergency management. The simulator also supports offline execution through rule-based fallbacks, making it suitable for academic research, training, and disaster response simulations.
Hemanth S, Manoj M, Harisha S, Dr Manjunath B· International Journal of Adv...· 0 citations
Natural hazards such as earthquakes can cause significant loss of life and extensive damage worldwide each year. Search and Rescue (SAR) operations in such critical situations require an urgent need for speed and efficiency, but are also complex challenges, given their dynamic and uncertain nature. This challenge requires approaches that integrate realistic environmental modeling with intelligent decision-making. Geospatial information systems are a powerful tool that can provide robust representations of complex terrains under time-critical circumstances, and artificial intelligence techniques can simplify decision-making and optimize decision-making under time-sensitive circumstances. In addition, context-aware SAR enables the precise identification and quantification of situational parameters, allowing better tailoring of operations to the real world. A context-aware SAR framework was developed in this study to identify and estimate key parameters in rescue operations, activities, and environmental conditions, and to model the problem within a spatial information system. And by integrating ant colony optimization (ACO), which is well-suited to modeling cooperative search behaviors, with the SARSA reinforcement learning algorithm, which is well-suited to discrete decision-making environments, a hybrid method was proposed. As part of this integration, a pheromone-like reinforcement mechanism was utilized to enhance overall performance and adaptability. A comparison of this hybrid approach against a standalone ACO model revealed that it outperformed it in both operational efficiency and solution quality, underscoring its potential for real-world SAR applications.
N. Bahrami, Meysam Argany, A. D. Boloorani et al.· International Journal of Eng...· 0 citations
Natural disasters with tremendous impacts on human lives, infrastructure, and ecosystems are frequent all over the world, which calls for intelligent, data-driven decision support systems for early diagnosis and effective crisis management. This paper demonstrates a Natural Disaster Diagnosis and Crisis Management System design and development featuring real-time environmental sensing, technological monitoring, and pre-emptive response planning within a unified decision-support framework. The proposed system include predict disasters using machine learning. It follows a modular architecture that integrates analytical risk assessment, formulation of preventive strategy, and dynamic action planning through an interactive GUI. NDDCMS structures disaster management into five operational phases, namely disaster diagnosis, early warning indicators, response during the disaster, post-disaster recovery, and preventive planning. Each phase integrates stepwise risk indicators and decision inputs to support officials and community response teams. It relies on quantifiable environmental parameters (e.g., precipitation, soil moisture, wind speed, and temperature fluctuation) and corresponding technological sensing mechanisms for the classification of risk levels and the improvement of early warning reliability. By focusing on a user-centered design and data-driven workflow, the system advances situational awareness, hastens decision-making, and closes the gap between disaster prediction and effective response. This framework enhances national and local-level disaster resilience by supporting viable risk reduction and crisis management strategies.
EL-Alfy A.E, Esmat Mona, Sakr Hagar· International Journal of Sci...· 0 citations
When a large disaster strikes, responders need a map of which buildings are damaged within hours. The models that do well on public benchmarks assume matched before-and-after imagery and a training set drawn from similar past events, and neither is usually available for a new disaster in its first day. We present 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. HASTE implements two methods that share one interface. The first requires the user to label polygons over the post-disaster scene, trains a small semantic segmentation model on that single scene, runs it over the whole image, and joins the per-pixel output to existing building footprints. The second embeds every footprint with a pretrained vision model, requires the user to label a handful of buildings, and fits a logistic regression in the browser that scores the rest of the scene in seconds. We describe the platform, both methods, and the engineering that supports them. We also report preliminary experiments on xBD showing that foundation-model embeddings pooled over footprints separate damaged from intact buildings using post-disaster imagery alone, matching a fully supervised ResNet-50 baseline with a twentieth of its labels. HASTE and its predecessors have supported more than thirty real-world disaster responses since 2023, spanning earthquakes, hurricanes, cyclones, floods, wildfires, and tornadoes, delivering results to humanitarian partners within hours to days of imagery becoming available. We close with the directions we think are most promising, including vision-language assessment, active learning, and damage models for roads and other infrastructure. HASTE is open source at https://github.com/microsoft/haste.
Caleb Robinson, Anthony Ortiz, S. Nsutezo et al.· 0 citations
To manage complex disaster risks, it is important to develop a response that spans multiple areas of expertise. Bridging the gap between law, social sciences, and natural sciences is an important part of any disaster risk reduction. It helps systems adapt quickly as AI technologies are constantly changing and have significantly impacted where law and the natural environment intersect, influencing legal systems and environmental policies, and how legal and environmental issues can prevent AI from having a greater impact on society and the economy. As part of a participatory assessment of production, with the assistance of legal experts, social and environmentalists, the key principles of responsible data analysis are proposed. These principles focus on security, transparency, fairness, accountability, and the ability to challenge or challenge decisions. This conversation describes how different disciplines can work together to create flexible legal systems that use AI, while drawing knowledge from the environmental and social sciences. The way environmentalists and decision-makers talk about useful and accurate information leads to differences that make the use of artificial intelligence difficult due to legal issues related to the reliability, reliability, and contentiousness of disaster management systems. If social media is useful for disaster risk reduction through AI, it's important to consider legal issues related to who is responsible and how sensitive the information used in disaster management is. Ideas for a fair and responsible process focus on the environment and encourage discussions on social and economic issues related to public participation. AI is also very important in education, as it brings together the next generation of law, social and natural sciences to jointly find solutions that bring together different disciplines in a balanced way. AI tools can be very useful in emergency management, but it's important to use them fairly and responsibly. You need to think about things like possible unfair benefits, clear explanations, and protecting people's personal data. Careful use of AI techniques, considering how AI, laws, and environmental risks interact with each other, helps create equitable and sustainable ways to collect and use data.
R. Paper, Research Supervisor Prof, Dominique Ferraro· The social science· 0 citations