Masters Theses

Keywords and Phrases

Altimeter Model; Digital Elevation Model; Kalman Filter; Navigation; Sensor Model Fidelity; Unscented Transform

Abstract

"Navigation filters used during an entry, descent, and landing scenario often receive measurements related to the terrain over which the vehicle is traversing. Models of the terrain can be presented in varying degrees of fidelity, providing increasingly realistic and accurate models of the surface. While it is ideal to provide the filter with the most realistic model possible, the performance is limited by the strict nature of the real time scenario. Therefore, it is of interest to determine what level of fidelity is necessary to achieve an accurate filtering solution. Determining how best to incorporate terrain-related data into navigation solutions can help produce the most robust and efficient navigation filter possible.

In the interest of determining a nominal filtering performance, an ellipsoidal model of the surface is used to create a baseline for the navigation filter. Increasingly complex digital elevation models are then utilized to better represent the surface over which the filter is navigating. These different fidelity models are then incorporated into various components of the filter to determine effects on the navigation performance. Simulations are performed that indicate that, as higher fidelity models are used, an increase in tuning noise must also be implemented to counteract the increasingly erratic nature of the terrain. The effects introduced by increasing terrain fidelity are also found to be reduced via the introduction of a lower fidelity estimate to the higher fidelity truth model.

Once simulations using an extended Kalman filter update are analyzed, alternative methods are developed to increase filter performance. Simulations with reduced uncertainties are performed to determine the effects of uncertainty on the terrain-based filtering solution. Following this, nonlinear transformations are implemented in an attempt to better account for the erratic nature of terrain-based models. While at lower fidelities these results provide increased performance, it is found that, as model fidelity is increased, the use of tuning noise or a lower fidelity estimate is once again required for a stable filtering solution"--Abstract, page iii.

Advisor(s)

DeMars, Kyle J.

Committee Member(s)

Pernicka, Hank
Hosder, Serhat

Department(s)

Mechanical and Aerospace Engineering

Degree Name

M.S. in Aerospace Engineering

Publisher

Missouri University of Science and Technology

Publication Date

Summer 2018

Pagination

xi, 98 pages

Note about bibliography

Includes bibliographic references (pages 95-97).

Rights

© 2018 Kenneth Michael Kratzer, All rights reserved.

Document Type

Thesis - Open Access

File Type

text

Language

English

Thesis Number

T 11382

Electronic OCLC #

1051222581

Share

 
COinS