Asynchronous multi-sensor target tracking with nonlinear models

Abstract

This thesis studies the target tracking problems with nonlinear dynamic models, under asynchronous, multi-rate, and multi-sensor conditions. Two sets of algorithms are proposed in this study. The first one is an algorithm for a tracking filter that is based on a nonlinear flight dynamics-based model. The second one deals with the estimation fusion for an asynchronous multi-rate multi-sensor environment using nonlinear models. The objectives of these algorithms are to allow for more comprehensive and various dynamic modes and to reflect a more realistic sensor environment than those that have been previously studied. The air traffic surveillance and control system is a typical example that can readily adopt the present algorithms because it processes the observations or the local state estimates from multiple asynchronous sensors, which detect targets with various flight modes. Firstly, a nonlinear aircraft tracking filter that uses a point mass flight dynamics model (PMFD) with three degrees of freedom is presented. Whereas the models used by the conventional air traffic control tracking filters are based on simple kinematics, the model for the present filter is based not only on the kinematic relations but also on the three-dimensional aircraft translational force equations and control variables. This allows for practical and sophisticated implementation of the attitude effects on translational acceleration. The control variables, which consist of the angle of attack, roll angle and thrust setting, are treated as states with random processes. Tracking with simulation data indicates that the present filter is superior to other single model-based filters and multiple model-based filters in terms of accuracy in position and course. Furthermore the model associated with it is insensitive to the flight motion types and design parameters. The results of tracking with real flight data also correspond well with those found by tracking with the simulation data. Secon...