Date of Award
Master of Science in Electrical Engineering
Department of Electrical and Computer Engineering
Michael J. Veth, PhD
Navigation performance in small unmanned aerial vehicles (UAVs) is adversely affected by limitations in current sensor technology for small, lightweight sensors. Because most UAVs are equipped with cameras for mission-related purposes, it is advantageous to utilize the camera to improve the navigation solution. This research improves navigation by matching camera images to a priori georegistered image data and combining this update with existing image-aided navigation technology. The georegistration matching is done by projecting the images into the same plane, extracting features using the techniques Scale Invariant Feature Transform (SIFT)  and Speeded-Up Robust Features (SURF) . The features are matched using the Random Scale and Consensus (RANSAC)  algorithm, which generates a model to transform feature locations from one image to another. In addition to matching the image taken by the UAV to the stored images, the effect of matching the images after transforming one to the perspective of the other is investigated. One of the chief advantages of this method is the ability to provide both an absolute position and attitude update. Test results using 15 minutes of aerial video footage at altitudes ranging from 1000m to 1500m demonstrated that transforming the image data from one perspective to the other yields an improvement in performance. The best system configuration uses SIFT on an image that was transformed into the satellite perspective and matched to satellite map data. This process is able to achieve attitude errors on the order of milliradians, and position errors on the order of a few meters vertically. The along track, cross track, and heading errors are higher than expected. Further work is needed on reliability. Once this is accomplished, it should improve the navigation solution of an aircraft, or even provide navigation grade position and attitude estimates in a GPS denied environment.
DTIC Accession Number
Webber, Frederick C., "Precision Navigation Using Pre-Georegistered Map Data" (2009). Theses and Dissertations. 2489.