Sensor planning for elusive targets

Problems such as searching for enemy units on a battlefield, detecting smugglers as they cross an international border, and skip tracing involve generating plans to employ limited collection resources to search for moving targets that are trying to avoid detection. This paper presents an automated approach for generating plans to search for ''elusive agents''. We also describe an implementation for the military problem of searching for mobile enemy ground units. The approach consists of three steps. First, it uses automated mobility and terrain analysis to hypothesize a set of possible movement plans for the targets. These plans are weighted with user-specified and heuristic probability estimates. Next, models of the available sensor resources are applied to identify observation ''windows''. These windows are regions in space and time where the target agents may be detected if they are following one of the hypothesized plans. Third, we generate a search plan for the available sensor assets (which can be any combination of mobile and fixed sensors) by heuristically searching through alternative subsets of the observation windows. Each search plan, defined as a temporally-ordered set of observation windows, is evaluated by exercising an automatically-constructed Bayesian network that summarizes the results of the terrain, route planning, and sensor coverage analysis. An empirical evaluation of this system was performed with results supporting its utility.