An epistemic utility based set-valued multiple-target tracker

The performance of a multiple target tracking method should degrade gracefully as the conditions of the collection become less favorable to optimal operation. The authors develop a multiple target trajectory-data association decision rule that is expressed in terms of two probabilities: one governing the informational value of the association hypothesis, and one governing the subjective belief, or credal, probability of the association hypothesis. A criterion of serious possibility is defined, and all trajectory-data associations that are seriously possible are retained; it is not necessary to resolve all conflicting hypotheses before processing more data. By stressing the avoidance, rather than the explicit minimization, of error, a decision rule is obtained for trajectory-data association that does not require the resolution of all conflicting hypotheses when the database does not contain sufficient information to do so reliably.