|dc.description.abstract||The visual feature-based Terrain-Aided Navigation (TAN) system presented in this thesis addresses the problem of constraining inertial drift introduced into the location estimate of Unmanned Aerial Vehicles (UAVs) in GPS-denied environment. The presented TAN system utilises salient visual features representing semantic or human-interpretable objects (roads, forest and water boundaries) from onboard aerial imagery and associates them to a database of reference features created a-priori, through application of the same feature detection algorithms to satellite imagery. Correlation of the detected features with the reference features via a series of the robust data association steps allows a localisation solution to be achieved with a finite absolute bound precision defined by the certainty of the reference dataset.
The feature-based Visual Navigation System (VNS) presented in this thesis was originally developed for a navigation application using simulated multi-year satellite image datasets. The extension of the system application into the mapping domain, in turn, has been based on the real (not simulated) flight data and imagery. In the mapping study the full potential of the system, being a versatile tool for enhancing the accuracy of the information derived from the aerial imagery has been demonstrated. Not only have the visual features, such as road networks, shorelines and water bodies, been used to obtain a position ’fix’, they have also been used in reverse for accurate mapping of vehicles detected on the roads into an inertial space with improved precision. Combined correction of the geo-coding errors and improved aircraft localisation formed a robust solution to the defense mapping application.
A system of the proposed design will provide a complete independent navigation solution to an autonomous UAV and additionally give it object tracking capability.||en_AU|
|dc.publisher||University of Sydney||en_AU|
|dc.publisher||Faculty of Engineering and IT||en_AU|
|dc.publisher||School of Aerospace, Mechanical and Mechatronic||en_AU|
|dc.rights||The author retains copyright of this thesis. It may only be used for the purposes of research and study. It must not be used for any other purposes and may not be transmitted or shared with others without prior permission.||en_AU|
|dc.subject.other||POST DG EXPORT SUBMISSION||en_AU|
|dc.title||Robust Modular Feature-Based Terrain-Aided Visual Navigation and Mapping||en_AU|
|dc.type.pubtype||Doctor of Philosophy Ph.D.||en_AU|
|Appears in Collections:||Sydney Digital Theses (Open Access)|