Sensor management with non-ideal sensor dynamics

Abstract – We describe a theoretically foundational but potentially practical control-theoretic basis for multisensormultitarget sensor management using a comprehensive, intuitive, system-level Bayesian paradigm based on random set theory. We focus on mobile sensors whose states are observed indirectly by internal actuator sensors. We determine optimal controls (future sensor states) using a "probabilistically natural ” sensor management objective function, the posterior expected number of targets (PENT). PENT is constructed using a new “maxi-PIMS ” optimization strategy to hedge against unknowable future observation-collections. It is used in conjunction with the PHD or MHC approximate multitarget filters