Integrated Multiple Hypothesis Nonlinear Tracking

"This proposal focuses on bringing together the NodeStar tracker (developed by METRON) and the Adaptive Sensor Fusion Multihypothesis Tracker (ASF-MHT) (developed by ORINCON) in a way that takes advantage of the best features of each to produce the mosteffective and practical system for data fusion and tracking. Significant research issues must be addressed to accomplish this goal. We will address these issues during the Phase I effort by developing an integrated MHT, Gaussian sum EKF, and Monte Carloparticle tracker that will serve as a testbed for research and development. We will develop and evaluate multiple approaches to approximating nonlinear likelihood functions with Gaussian sums. We will develop Gaussian sum conversion techniques thatdetermine the number of terms in the Gaussian sum, the location of the means, and the covariances, and using the integrated testbed, we will compare the accuracy and speed of the Gaussian sum conversion techniques with the accuracy and speed of the MonteCarlo particle tracker. We will compare Gaussian sum extended Kalman filtering results to Monte Carlo particle filter results over an initial set of scenarios. We will compare the accuracy of posterior distributions and computation time. We expect thatthis effort will result in practical implementation of a superior data fusion and tracking system that will be the standard against which other such systems will be compared. It will provide th