Date of Award

3-18-2009

Document Type

Thesis

Degree Name

Master of Science in Electrical Engineering

Department

Department of Electrical and Computer Engineering

First Advisor

John F. Raquet, PhD

Abstract

Determining the position of team members is always useful information, whether it is a team of firefighters fighting a blaze or combatants clearing a building in the field. This information becomes even more decisive for the people responsible for their safety. To accomplish this in areas denied Global Navigation Satellite System (GNSS), such as around buildings or in steep valleys, alternative methods must be used. Radio ranging systems have been a part of the navigation solution for years. They unfortunately have poor performance in certain areas, such as inside buildings, due to multipath and other errors. To improve the position estimate it is believed that using vision, consumer grade inertial navigation systems, and any other measurement available can aid the navigation solution. To accomplish this, an extended Kalman filter was developed. It was configured as a centralized filter. This produced a baseline, showing that as image measurements were added, the navigation solution did improve. To simulate this with multiple vehicles and/or soldiers required a large state vector for the Kalman filter. To manage the large number of states and efficiently incorporate them into influence matrices, a "Rosetta stone" was designed for state management. This "Rosetta stone" breaks the states into simpler blocks such as position and attitude for the soldier and position for the image features. This in turn made updating the influence matrix and covariance matrix a smoother process. The impact of adding image measurements has been two fold. First, the position RMS errors were reduced by approximately a factor of 2. Second, the attitude which fluctuated greatly in the radio only cases was reduced by a factor of 10 through image aiding.

AFIT Designator

AFIT-GE-ENG-09-15

DTIC Accession Number

ADA499625

Download

Share

COinS