Fusion using Target Motion Analysis and Geographic Data for Maritime Applications

This work presents a study that uses a fusion algorithm to improve the accuracy of tracking estimates employing target motion analysis and geographic data. The target tracking process is performed by an unmanned aerial vehicle that has an onboard bearings-only sensor to perform a surveillance mission. Geographic information comes from GIS and databases of real marine vessels. The used target tracking algorithm is a Course Constrained Interactive Multiple Model (IMM), able to detect on-course and off-course modes and track using a course constrained filter and a conventional filter respectively. The performed simulations results present the case of a vessel approaching a port and finally entering a harbour. Final results consist on the comparison of the tracking results using a conventional Extended Kalman Filter and the Course Constrained IMM in two different cases; the first one is a semi-ideal case, where the vessel is on course most of the time, and the second case uses actual vessel data which is off course most of the time.