{"id":{"repo_id":"uts","oai_identifier":"oai:opus.lib.uts.edu.au:10453/140553"},"canonical_url":"https://search.dev.ndltd.org/etd/uts/oai:opus.lib.uts.edu.au:10453/140553","repository":{"repo_id":"uts","name":"University of Technology Sydney","base_url":"https://opus.lib.uts.edu.au/oai/request"},"display":{"title":"Towards Observable Urban Visual SLAM","abstract":"Visual Simultaneous Localisation and Mapping (V-SLAM) is the subject of robot state and environment map estimation by drawing inference on camera captured data. It has been a major branch of research and popular in application owing to the rich information and low cost in vision measurement acquisition. However, for applications in urban environments, where the camera-mounted vehicle moves along a straight line direction towards the road scene, a large number of features suffer difficulty in depth estimation due to their small parallax angles, as a result the classical V-SLAM algorithm encounters instability and the system state is often unobservable. This thesis addresses the issue of Urban SLAM observability associated with monocular cameras. It proposes a novel Bundle Adjustment (BA) formulation that addresses the problem from a fundamental approach – by parameterising map points in an on-manifold ray parallax form the SLAM formulation has a stable configuration that guarantees local state observability despite of presence of low parallax features. V-SLAM is known to be highly non-convex from its projective image formation principle. Slight off-optimal initial values easily lead to sub-optimal final state estimates. In Urban SLAM this is further exacerbated by collinear camera motion that causes ambiguity in initial state estimation. A robust initialisation method is proposed in this thesis to provide unique near-optimal initial estimates effectively addressing collinearity issues. For practical use of our algorithm, we demonstrate how the urban scene friendly V-SLAM algorithms are integrated into a real-time Visual Inertial Navigation system (VINS). A series of quantitative analyses are performed on a few benchmark datasets, demonstrating effectiveness of our algorithm in urban environments.","abstract_html":"Visual Simultaneous Localisation and Mapping (V-SLAM) is the subject of robot state and environment map estimation by drawing inference on camera captured data. It has been a major branch of research and popular in application owing to the rich information and low cost in vision measurement acquisition. However, for applications in urban environments, where the camera-mounted vehicle moves along a straight line direction towards the road scene, a large number of features suffer difficulty in depth estimation due to their small parallax angles, as a result the classical V-SLAM algorithm encounters instability and the system state is often unobservable. This thesis addresses the issue of Urban SLAM observability associated with monocular cameras. It proposes a novel Bundle Adjustment (BA) formulation that addresses the problem from a fundamental approach – by parameterising map points in an on-manifold ray parallax form the SLAM formulation has a stable configuration that guarantees local state observability despite of presence of low parallax features. V-SLAM is known to be highly non-convex from its projective image formation principle. Slight off-optimal initial values easily lead to sub-optimal final state estimates. In Urban SLAM this is further exacerbated by collinear camera motion that causes ambiguity in initial state estimation. A robust initialisation method is proposed in this thesis to provide unique near-optimal initial estimates effectively addressing collinearity issues. For practical use of our algorithm, we demonstrate how the urban scene friendly V-SLAM algorithms are integrated into a real-time Visual Inertial Navigation system (VINS). A series of quantitative analyses are performed on a few benchmark datasets, demonstrating effectiveness of our algorithm in urban environments.","abstract_has_math":false,"creators":["Liu, Li Yang"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2020,"date_issued":"2020","date_published":"2020","updated_at":"2026-07-24T06:32:35Z","subjects":[],"languages":["en_US"],"rights":["info:eu-repo/semantics/openAccess","The author owns the copyright in this thesis including all reproduction and reuse rights for the work. The work may not be altered without the permission of the copyright owner. Attribution is essential when quoting or paraphrasing from this thesis.","au.edu.uts.lib/ppc"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/10453/140553","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Liu, Li Yang"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2020-05-08T00:40:25Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2020-05-08T00:40:25Z"]},{"key":"dc:date.issued","label":"Date","values":["2020"]},{"key":"dc:relation","label":"Dc Relation","values":["https://opus.lib.uts.edu.au/bitstream/10453/140553/2/02whole.pdf"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en_US"]},{"key":"dc:rights","label":"Dc Rights","values":["info:eu-repo/semantics/openAccess","The author owns the copyright in this thesis including all reproduction and reuse rights for the work. The work may not be altered without the permission of the copyright owner. Attribution is essential when quoting or paraphrasing from this thesis.","au.edu.uts.lib/ppc"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["http://hdl.handle.net/10453/140553"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["University of Technology Sydney. Faculty of Engineering and Information Technology."]},{"key":"dc:description.abstract","label":"Abstract","values":["Visual Simultaneous Localisation and Mapping (V-SLAM) is the subject of robot state and environment map estimation by drawing inference on camera captured data. It has been a major branch of research and popular in application owing to the rich information and low cost in vision measurement acquisition. However, for applications in urban environments, where the camera-mounted vehicle moves along a straight line direction towards the road scene, a large number of features suffer difficulty in depth estimation due to their small parallax angles, as a result the classical V-SLAM algorithm encounters instability and the system state is often unobservable. This thesis addresses the issue of Urban SLAM observability associated with monocular cameras. It proposes a novel Bundle Adjustment (BA) formulation that addresses the problem from a fundamental approach – by parameterising map points in an on-manifold ray parallax form the SLAM formulation has a stable configuration that guarantees local state observability despite of presence of low parallax features. V-SLAM is known to be highly non-convex from its projective image formation principle. Slight off-optimal initial values easily lead to sub-optimal final state estimates. In Urban SLAM this is further exacerbated by collinear camera motion that causes ambiguity in initial state estimation. A robust initialisation method is proposed in this thesis to provide unique near-optimal initial estimates effectively addressing collinearity issues. For practical use of our algorithm, we demonstrate how the urban scene friendly V-SLAM algorithms are integrated into a real-time Visual Inertial Navigation system (VINS). A series of quantitative analyses are performed on a few benchmark datasets, demonstrating effectiveness of our algorithm in urban environments."]},{"key":"dc:format","label":"Dc Format","values":["Thesis (PhD)"]},{"key":"dc:title","label":"Title","values":["Towards Observable Urban Visual SLAM"]}]}],"canonical_facts":{"dc:creator":["Liu, Li Yang"],"dc:date.accessioned":["2020-05-08T00:40:25Z"],"dc:date.available":["2020-05-08T00:40:25Z"],"dc:date.issued":["2020"],"dc:description":["University of Technology Sydney. Faculty of Engineering and Information Technology."],"dc:description.abstract":["Visual Simultaneous Localisation and Mapping (V-SLAM) is the subject of robot state and environment map estimation by drawing inference on camera captured data. It has been a major branch of research and popular in application owing to the rich information and low cost in vision measurement acquisition. However, for applications in urban environments, where the camera-mounted vehicle moves along a straight line direction towards the road scene, a large number of features suffer difficulty in depth estimation due to their small parallax angles, as a result the classical V-SLAM algorithm encounters instability and the system state is often unobservable. This thesis addresses the issue of Urban SLAM observability associated with monocular cameras. It proposes a novel Bundle Adjustment (BA) formulation that addresses the problem from a fundamental approach – by parameterising map points in an on-manifold ray parallax form the SLAM formulation has a stable configuration that guarantees local state observability despite of presence of low parallax features. V-SLAM is known to be highly non-convex from its projective image formation principle. Slight off-optimal initial values easily lead to sub-optimal final state estimates. In Urban SLAM this is further exacerbated by collinear camera motion that causes ambiguity in initial state estimation. A robust initialisation method is proposed in this thesis to provide unique near-optimal initial estimates effectively addressing collinearity issues. For practical use of our algorithm, we demonstrate how the urban scene friendly V-SLAM algorithms are integrated into a real-time Visual Inertial Navigation system (VINS). A series of quantitative analyses are performed on a few benchmark datasets, demonstrating effectiveness of our algorithm in urban environments."],"dc:format":["Thesis (PhD)"],"dc:identifier.uri":["http://hdl.handle.net/10453/140553"],"dc:language.iso":["en_US"],"dc:relation":["https://opus.lib.uts.edu.au/bitstream/10453/140553/2/02whole.pdf"],"dc:rights":["info:eu-repo/semantics/openAccess","The author owns the copyright in this thesis including all reproduction and reuse rights for the work. The work may not be altered without the permission of the copyright owner. Attribution is essential when quoting or paraphrasing from this thesis.","au.edu.uts.lib/ppc"],"dc:title":["Towards Observable Urban Visual SLAM"],"dc:type":["Thesis"]},"updated_at":"2026-07-24T06:32:35Z"}