Incremental fusion of GMTI tracks for road map estimation

Vector roadmaps are invaluable for various applications such as navigation and tracking. A new algorithm is presented aimed at incrementally producing exact, up-to-date vector roadmaps using GMTI tracks. In most existing work the track points are processed independently. Track points are individually associated with segments of existing roads and the road segment locations are corrected accordingly. In this work, rather than being viewed as sequences of points, the tracks and roads are considered as unified entities and are therefore fused as a whole. This paper focuses on the heart of the algorithm, namely, on the mathematics of curve to curve fusion whereby an incoming track curve is fused with an existing road curve to obtain a new, more accurate, estimate of the road trajectory. A dynamical programming method is employed to align the curves and a weighted average curve is then computed. The results of a simulated application are presented.