SBIR Ph2 FAST

This proposal addresses the challenge of incorporating non-kinematic features into tactical active sonar tracking algorithms applied to surface anti-submarine warfare. The central focus is the development of three general methods that use feature information (i.e., their statistical characterization) to improve both data association and filtering. The methods are general in that they apply to a wide variety of features and trackers. They will be implemented and tested in STR’s Multiple Hypothesis Tracker (MHT). Two avenues to expand the core capability of the feature-aided tracking techniques are also proposed. The first explores multi-stage MHT architectures that fuse kinematic and feature information from multiple sensors. The second avenue explores algorithms to auto-tune the tracker to avoid feature model mismatch. Based on gains seen in other domains, these feature-aided tracking techniques and architectures are expected to improve multi-target tracking performance in the presence of heavy clutter and high ambient noise.