A pursuit and evasion problem with measurement uncertainty (Q2535950)

A pursuit-evasion process with uncertain state-dependent measurements, in finite, discrete time and in a finite, discrete state space, is considered. Three types of strategies which might be employed in such a process are compared, and attention is concentrated on the behavior strategy, the least well known of the three types, but often the simplest optimal strategy to employ. A variation of the Brown-Robinson fictitious play algorithm is presented which can be used to compute behavior strategies in the case of perfect recall processes. Two examples are given in which optimal behavior strategies are computed using the algorithm and compared with a type of plausible but nonoptimal separation strategy.