Advance Tracking Algorithms to Meet Modern Threats - MP 14-14

Modern UAVs and fifth-generation aircraft are capable of supermaneuverability that includes rapid changes in acceleration, high-G turns, and rapid changes in radar cross-section (RCS) and/or low RCS profile. These traits pose problems to conventional Kalman-based trackers (including the extended Kalman filter (EKF) and unscented Kalman filter (UKF)) which assume linearized models with additive Gaussian noise processes and require data be processed to provide"detections."Consequently, systems that use these trackers will break lock. In contrast, grid and particle filter (PF) trackers allow nonlinear state and measurement models, do not require additive Gaussian noise processes, and can use track-before-detect (TBD) measurement models to achieve significantly improved performance. In Phase I, we propose to evaluate several nonlinear filtering (NLF) algorithms to determine their performance in terms of tracking ability and computational requirements. The simulation study will include targets that exhibit supermaneuverability, low RCS, specular scintillation properties, and combinations of the preceding. The techniques we will investigate include the EKF, UKF, Grid Filter, Resampling PF, and Homotopy PF with multiple motion models. We will report performance/compute tradeoffs for the algorithms and identify capability gaps to be addressed in Phase II.