Evaluation of Data Association and Fusion Algorithms for Tracking in the Presence of Measurement Loss
Vps Naidu, G Girija, Jitendra R. Raol · AIAA Guidance, Navigation, and Control Conference and Exhibit · 2003
Tracking in multi sensor multi target (MSMT) scenario is a complex problem due to the uncertainties in the origin of observations. Solution to this problem requires appropriate gating and data association procedures to associate measurements with targets. A PC MATLAB program based on track-oriented approach is evaluated which uses nearest neighbor Kalman filter (NNKF) and probabilistic data association filter (PDAF) for tracking multiple targets from data of multiple sensors. For track-to-track fusion, state vector fusion philosophy is employed. The tracking performance in the presence of simulated track loss and recovery as well as in clutter is evaluated. During data loss PDAF performed better than NNKF. In the presence of mild clutter and sparse target scenarios, the NNKF and PDAF give similar performance.