Adaptive search for sparse targets with informative priors

This works considers the problem of energy constrained adaptive search for sparse targets given probabilistic prior knowledge of target locations. An Adaptive Resource Allocation Policy (ARAP) was introduced by Bashan (2008), showing significant gains over standard methods can be achieved without prior knowledge on the targets´ locations. This work extends ARAP to account for nonuniform prior knowledge. It is shown that potential gains exist as compared to ARAP. Moreover, we show that by overestimating the true region of interest, the proposed search policy can always outperform ARAP in terms of worst-case gain. Lastly, results from an application involving estimating the approach of airplanes at an airport suggest that bi-level piecewise uniform priors are adequate approximations.