Discrete search allocation with object uncertainty

We develop a new approach for assigning the optimal allocation of multiple searchers/sensors to a discrete set of search cells to find a hidden object. In contrast to existing formulations of the discrete search problem, we consider the situation in which the object is of uncertain type. Such formulations are practical for problems where a group of searchers is sent to look for an object of uncertain disposition, which is commonplace in search and rescue as well as many military search applications. We formulate a new mathematical model for this problem in which the object uncertainty is accounted for as an additive perturbation to the traditional Bayesian formulation of discrete search. This modelling approach is then proven to be amenable to optimisation with a greedy algorithm. Numerical examples illustrate the improved search performance gained from these improved allocations.