Multiple Target Tracking UsingMaximumLikelihood Probabilistic DataAssociation

TheMaximumLikelihood-Probabilistic Data Association (MLPDA)target tracking algorithm isef- fective intracking verylowobservable targets. A key limitation ofMLPDA isthatitisrestricted totracking asingle target. We derive andimplement amultiple tar- getversion ofMLPDA called Joint MLPDA (JMLPDA). WhiletheJMLPDA implementation presented inthis paperisfocused onatwo-target case, thisalgorithm isextensible toanynumberoftargets. TheMLPDA andJMLPDAalgorithms arecombined toformamulti- target MLPDA tracking algorithm. Performance ofthe JMLPDAandthemulti-target MLPDA algorithms are compared toa Probabilistic Multi-Hypothesis Tracker (PMHT)fortwocrossing targets, focusing ontrack man- agement/update. Simulation results showthatunder conditions ofheavyclutter, themulti-target MLPDA outperforms PMHT intermsofreduced track errors and longer track life.