Recursive Filtering Algorithms for Ship Tracking
Warren W. Willman · 1976
Some recursive filtering algorithms were developed for tracking ships when observations are sporadic and imprecise. Tracking with position-only observations was emphasized, but a procedure was also developed for utilizing possible independent observations of ship velocity. Two basic algorithms are considered: a Kalman filter with adaptive driving noise for generating estimates (and containment ellipses) for current and future ship positions, and a corresponding Bayesian smoother for generating estimates of past positions. The driving noise was treated as a velocity term in a continuous-time model of ship's motion. The details of these two algorithms were developed for tracking on a plane, on a sphere in geographical coordinates, and on a sphere in three-dimensional rectilinear coordinates. A FORTRAN implementation and some corresponding numerical results were developed for the planar case.