Soroush SamaniWhen severe flooding threatens a town, emergency services must help people reach safety. But even a carefully prepared evacuation plan can quickly become outdated. Roads may flood, traffic queues can grow, and drivers may not follow recommended routes. Emergency teams must make difficult decisions, often with very little time.
This research explores how artificial intelligence (AI) and mathematical optimisation could help emergency services respond as conditions change.
The project will develop and test a decision-support system combining three elements. First, a computer simulation will recreate changing flood and traffic conditions, using real geographical data from Cockermouth, a flood-prone town in England. Second, a mathematical model will investigate possible changes to evacuation plans, taking account of road capacity, traffic congestion and uncertain flood developments. Third, an AI agent will bring together information from flood forecasts, traffic data and reports from emergency responders, both generated synthetically. It will use the mathematical model and simulation tools to investigate alternative actions and present recommendations.
Importantly, AI will not make the final decisions. Human emergency operators will review the evidence and decide what action to take.
The research will examine whether this approach can produce reliable recommendations quickly enough to be useful. Its findings could help emergency planners explore safer evacuation strategies and improve our understanding of how AI can support human decisions during emergencies.