Disaster management

Disaster management: This involves smart applications for environmental sustainability refers to the use of advanced technologies and data-driven approaches to prepare for, respond to, and recover from disasters while minimizing the environmental impact. 

Some smart applications for environment sustainability in disaster management include:

• Early Warning Systems: This involves the use of sensors and data analytics to detect natural disasters such as floods, earthquakes, and wildfires in real-time, providing advanced warning to affected populations.

• Disaster Response: This involves the use of drones, autonomous vehicles, and other technologies to assist in disaster response efforts, such as search and rescue operations, damage assessment, and the delivery of critical supplies.

• Energy Resilience: This involves the use of microgrids, renewable energy sources, and energy storage systems to provide reliable and sustainable energy in the aftermath of a disaster.

• Smart Infrastructure: This involves the use of sensors and advanced analytics to monitor and maintain critical infrastructure such as bridges, roads, and water treatment plants, improving their resilience to disasters.

• Waste Management: This involves the use of smart waste management systems to quickly and efficiently manage waste generated by disasters, reducing the environmental impact and promoting sustainable waste management practices.

Project Details:   An AI-Powered Multi-Agentic Platform for Human-Centered Disaster Response :  ResQConnect

Floods and landslides create complex response environments where emergency coordinators must make rapid decisions despite uncertain information, disrupted communication, and limited resources. Many current disaster-management systems rely on manual coordination and fragmented information sources, making it difficult to process citizen reports and allocate relief resources efficiently. Our final year research project introduces ResQConnect, a human-centered AI-assisted disaster response decision-support platform designed to improve coordination and situational awareness during hydrological disasters. The system integrates an agentic Retrieval-Augmented Generation (RAG) pipeline to convert unstructured help requests into structured task plans, an adaptive event-triggered multi-commodity resource distribution algorithm for efficient routing based on urgency and distance, and a lightweight on-device language model that provides offline triage and safety guidance when connectivity is limited. Additionally, an RL-optimized conversational assistant learns when external knowledge retrieval is necessary during multi-turn interactions, reducing unnecessary retrieval while maintaining response quality. ResQConnect functions strictly as a decision-support system where human coordinators review and approve all critical actions, and experimental evaluations using realistic disaster scenarios show improvements in task relevance, coordination efficiency, and overall system robustness compared to baseline RAG and static routing approaches.

Awards .... ResQConnect Project

SLAAI AI Project Award:   Best Undergraduate AI Project Award 

IntelliHack - Champion: UoM Newsletter - Page 8

SDG Sprints - Champion: IEEE Sri Lanka Section SIGHT - SDG Sprints

INNOVA - Champion: INNOVA held under TechX Sri Lanka 2025

SDG Youth Challenge - 2nd Runners-up

Project Team:

 Mr. Savinu Aththanayake; Ms. Chemini Mallikarachchi, Ms. Janeesha Wickramasinghe, Mr. Sajeev Kugarajah 

Supervisor: Prof. Dulani Meedeniya