The para-normal Bayes multi-target filter and the spooky effect

Abstract

The Probability Hypothesis Density (PHD) and Cardinalized PHD (CPHD) filters exhibit a counter intuitive behaviour called the ”spooky effect” where upon a missed detection, the PHD mass in the vicinity of the undetected target is shifted to the vicinity of the detected targets, regardless of the distance between the targets. This raises the question of whether spookiness is an artifact of the random finite set formulation of the multi-target filtering problem. Using a para-normal implementation of the labeled multi-target Bayes filter, we show that this filter does not exhibit the “spooky effect” observed in the PHD/CPHD filters.