Robust tracking of marginally observable targets

The authors address the robust determination of a target trajectory from acoustic or radar data under conditions that may render conventional estimation techniques unreliable, such as weak or fading signals, sensor failure, or poor signal collection geometries. A set-valued estimator that provides the set of all trajectories that are consistent with the observed signals and available contextual information is introduced. Simulation results that demonstrate the operation of the set-valued estimator are presented