Optimal team deployment in urban search and rescue

The problem of optimally deploying urban search and rescue (USAR) teams to disaster sites in post-disaster circumstances is formulated as a multistage stochastic program (MSP). A portion of sites requiring assistance arrive dynamically over the decision horizon and key problem characteristics are known only with uncertainty a priori. The problem seeks to identify a set of tours for USAR teams so as to maximize the total expected number of people that can be saved by attending to all or a subset of disaster sites within the disaster region. Decisions are taken dynamically over the decision horizon as situational awareness improves and survival likelihood diminishes with the aim of increasing the expected number of saved lives. To overcome the expensive computational effort associated with solving a MSP, a column generation-based strategy that consists of solving a series of interrelated two-stage stochastic programs with recourse within a shrinking time horizon is developed.

► Formulates the USAR Team Deployment Problem as a multistage stochastic program. ► Captures diminishing survival likelihood in maximizing lives saved. ► Considers probabilistic and dynamic nature of conditions following a disaster. ► Proposes dynamic solution framework for sequential stochastic information process. ► Exploits solutions from prior decision epochs within shrinking time horizon.