UC Santa Cruz

Algorithms based on traditional notion of tracking as a state estimation problem

yield just a single interpretation of the data. For some applications, the ability to

identify ambiguities and compare different interpretations using a well-defined measure

of confidence is critical. Such applications require a direct solution to the data association

problem in order to characterize the relevant uncertainty. This notion of tracking

has received relatively little attention largely due to a failure to recognize its utility

beyond maintaining the state estimation process. As a result, the options available to

the practitioner are limited and the performance of statistical data association models

is not well understood, especially in terms of the quality of the sample they produce.

This work has sought to change that by developing a new data association

model that extends the scope and flexibility of existing models. The questions of how

to specify an objective prior distribution over data association hypotheses and how

to efficiently perform inference on the high-dimensional posterior distribution are very

much open. To help provide answers, we considered numerous different priors, including

Bayesian nonparametric models and several models never before applied to tracking.

With regard to inference, we considered various implementations of Markov chain Monte

Carlo (MCMC) and population Monte Carlo (PMC) samplers. A comprehensive evaluation

was performed in the context of a wide-area radar surveillance application.