Master's Thesis Defense, A Multiple Target Tracker for Nonlinear Rotational Object Motion Using Range and Range-Rate Measurements

Abstract

In many applications, it is necessary to track an unknown number of targets with a remote sensor. Several applications for which radar is used to sense the environment include anti-aircraft warfare, tracking of space debris, and missile defense. These tracking algorithms generally consist of two major components: an association scheme and a filtering method. If the radar has a high range resolution resulting from large bandwidth, also called wideband radar, one can receive multiple returns from scatterers on a single target returns in from one pulse. In this case, the complexity of the association algorithm must increase as the number of measurements per object is unknown as well as the number of objects. Further complexities can also arise in the filtering method. If the target dynamics are non-linear functions of time than an adequate filter motion model should reflect this causing the state equations characterizing this motion to be non-linear and time- varying. The target measurements, as a function of this filter state, may also be non-linear. This paper outlines the development of such a tracker with both a non-linear object motion model and measurement equations for wideband radar. In the particular problem considered, it is necessary to track multiple scatterers on an unknown number of objects, whose motion includes a non-linear component: rotational velocity, in a two- dimensional frame. The measurements, from generic wideband radar, are scatterer range and range-rate (in analogy to Doppler) and are non-linear functions of scatterer position and velocity, components of the state parameters. The tracker design uses an Extended Kalman Filter to model the non-linear dynamics and measurement models and a multi-pronged association scheme to determine the scene and track its evolution over time.

Committee Members: Prof. Eric Miller, Prof. Joseph Noonan and Dr. Keh-Ping Dunn, Lincoln Laboratory