{"id":{"repo_id":"buffalo","oai_identifier":"oai:ubir.buffalo.edu:10477/79955"},"canonical_url":"https://search.dev.ndltd.org/etd/buffalo/oai:ubir.buffalo.edu:10477/79955","repository":{"repo_id":"buffalo","name":"Buffalo","base_url":"https://ubir.buffalo.edu/oai/request"},"display":{"title":"Adaptive Visual Slam for Indoor Environments","abstract":"Ph.D.","abstract_html":"Ph.D.","abstract_has_math":false,"creators":["Hashemifar, Zakieh Sadat; 0000-0003-2157-9448"],"institution":"State University of New York at Buffalo","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Dantu, Karthik","Computer Science and Engineering"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2019,"date_issued":"2019-07-30T15:11:23Z","date_published":"2019-07-30T15:11:23Z","updated_at":"2026-08-21T22:21:56Z","subjects":["robotics","computer science"],"languages":["eng"],"rights":["Users of works found in University at Buffalo Institutional Repository (UBIR) are responsible for identifying and contacting the copyright owner for permission to reuse. University at Buffalo Libraries do not manage rights for copyright-protected works and cannot assist with permissions.","Copyright retained by author."],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/10477/79955","outbound_label":"Handle","outbound_source":"dc:identifier"},"source_record":{"url":"https://ubir.buffalo.edu/oai/request?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Aubir.buffalo.edu%3A10477%2F79955","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Dantu, Karthik","Computer Science and Engineering"]},{"key":"dc:creator","label":"Author","values":["Hashemifar, Zakieh Sadat; 0000-0003-2157-9448"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2019-07-30T15:11:23Z","2019","2019-05-15 15:40:39"]},{"key":"dc:publisher","label":"Institution","values":["State University of New York at Buffalo"]},{"key":"dc:relation","label":"Dc Relation","values":["Supplementary file: Hashemifar-wifi-augmentation.mp4 (video), Wi-Fi augmentation of ORBSLAM for removing false loop closures"]},{"key":"dc:type","label":"Dc Type","values":["Text","Dissertation","Video"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["robotics","computer science"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Users of works found in University at Buffalo Institutional Repository (UBIR) are responsible for identifying and contacting the copyright owner for permission to reuse. University at Buffalo Libraries do not manage rights for copyright-protected works and cannot assist with permissions.","Copyright retained by author."]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/10477/79955"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Ph.D.","A fundamental building block for autonomous robots is Simultaneous Localization and Mapping (SLAM). Beyond robots, many wearables (such as smart glasses) provide augmented reality and virtual reality applications that could benefit from indoor localization. Different SLAM approaches accommodate a variety of sensors and computing requirements. However, recent trends have seen the use of regular and depth cameras together for mapping. Visual SLAM algorithms are exposed to several challenges including huge map size, perceptual aliasing and high computational cost. Furthermore, constant operation in the same environment for longer periods of time expose visual SLAM algorithms to a new challenge; semi-static and dynamic environment mapping.These challenges affect the accuracy, efficiency, and viability of visual SLAM algorithms. This thesis explores various improvements to SLAM for indoor use. The contributions of this thesis are threefold: First, we propose an algorithmic pipeline that attributes semantics to landmarks and reduces the dimensionality of the represented map greatly. Since manycommon urban scenes (such as offices, homes) contain cuboidal objects (such as cabinets, tables), we selected cuboids for object representation. We develop a metric to perform such attribution consistently so they can be used as landmarks for mapping/navigation. We have tested our pipeline on different datasets and show that we can reduce the map representation by more than 50% while maintaining localization accuracy. Second, we propose an algorithm for augmenting visual SLAM algorithms with Wi-Fi sensing. Wi-Fi routers are ubiquitous in most urban settings including offices, homes, malls, and airports. Most robots and computing devices such as cellphones and wearables are typically equipped with a Wi-Fi radio for communication. We propose a method to utilize Wi-Fi sensing to alleviate the aforementioned perceptual aliases and high computation time challenges. We integrate Wi-Fi similarity into three well-known SLAM algorithms (RGBD SLAM, RTAB-Map, and ORB-SLAM) and experimentally demonstrate such Wi-Fi augmentation can help improve these three visual SLAM algorithms using four datasets we collected. Overall, we show that our approach can improve the accuracy of visual SLAM algorithms by 11% on average and reduce computation time on average by 15% to 25%. Third, we incorporate persistence filters into ORB-SLAM, a well-known visual SLAM algorithm for handling semi-static and dynamic environments. We observe that simply integrating the previous proposal results in inefficient persistence reasoning. Through a series of modifications, we improve this persistence filtering. Using two locally collected datasets, we demonstrate the utility of such persistence filtering as well as our customizations in ORB-SLAM. Overall, incorporating persistence filtering could result in a significant reduction inmap size (about 30-40% in the best case) and a corresponding reduction in run-time while retaining similar accuracy to methods that use much larger maps.","**To request an accessible version of the file(s) associated with this item, contact library@buffalo.edu. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.**"]},{"key":"dc:format","label":"Dc Format","values":["application/pdf","video/mp4"]},{"key":"dc:title","label":"Title","values":["Adaptive Visual Slam for Indoor Environments"]}]}],"canonical_facts":{"dc:contributor":["Dantu, Karthik","Computer Science and Engineering"],"dc:creator":["Hashemifar, Zakieh Sadat; 0000-0003-2157-9448"],"dc:date":["2019-07-30T15:11:23Z","2019","2019-05-15 15:40:39"],"dc:description":["Ph.D.","A fundamental building block for autonomous robots is Simultaneous Localization and Mapping (SLAM). Beyond robots, many wearables (such as smart glasses) provide augmented reality and virtual reality applications that could benefit from indoor localization. Different SLAM approaches accommodate a variety of sensors and computing requirements. However, recent trends have seen the use of regular and depth cameras together for mapping. Visual SLAM algorithms are exposed to several challenges including huge map size, perceptual aliasing and high computational cost. Furthermore, constant operation in the same environment for longer periods of time expose visual SLAM algorithms to a new challenge; semi-static and dynamic environment mapping.These challenges affect the accuracy, efficiency, and viability of visual SLAM algorithms. This thesis explores various improvements to SLAM for indoor use. The contributions of this thesis are threefold: First, we propose an algorithmic pipeline that attributes semantics to landmarks and reduces the dimensionality of the represented map greatly. Since manycommon urban scenes (such as offices, homes) contain cuboidal objects (such as cabinets, tables), we selected cuboids for object representation. We develop a metric to perform such attribution consistently so they can be used as landmarks for mapping/navigation. We have tested our pipeline on different datasets and show that we can reduce the map representation by more than 50% while maintaining localization accuracy. Second, we propose an algorithm for augmenting visual SLAM algorithms with Wi-Fi sensing. Wi-Fi routers are ubiquitous in most urban settings including offices, homes, malls, and airports. Most robots and computing devices such as cellphones and wearables are typically equipped with a Wi-Fi radio for communication. We propose a method to utilize Wi-Fi sensing to alleviate the aforementioned perceptual aliases and high computation time challenges. We integrate Wi-Fi similarity into three well-known SLAM algorithms (RGBD SLAM, RTAB-Map, and ORB-SLAM) and experimentally demonstrate such Wi-Fi augmentation can help improve these three visual SLAM algorithms using four datasets we collected. Overall, we show that our approach can improve the accuracy of visual SLAM algorithms by 11% on average and reduce computation time on average by 15% to 25%. Third, we incorporate persistence filters into ORB-SLAM, a well-known visual SLAM algorithm for handling semi-static and dynamic environments. We observe that simply integrating the previous proposal results in inefficient persistence reasoning. Through a series of modifications, we improve this persistence filtering. Using two locally collected datasets, we demonstrate the utility of such persistence filtering as well as our customizations in ORB-SLAM. Overall, incorporating persistence filtering could result in a significant reduction inmap size (about 30-40% in the best case) and a corresponding reduction in run-time while retaining similar accuracy to methods that use much larger maps.","**To request an accessible version of the file(s) associated with this item, contact library@buffalo.edu. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.**"],"dc:format":["application/pdf","video/mp4"],"dc:identifier":["http://hdl.handle.net/10477/79955"],"dc:language":["eng"],"dc:publisher":["State University of New York at Buffalo"],"dc:relation":["Supplementary file: Hashemifar-wifi-augmentation.mp4 (video), Wi-Fi augmentation of ORBSLAM for removing false loop closures"],"dc:rights":["Users of works found in University at Buffalo Institutional Repository (UBIR) are responsible for identifying and contacting the copyright owner for permission to reuse. University at Buffalo Libraries do not manage rights for copyright-protected works and cannot assist with permissions.","Copyright retained by author."],"dc:subject":["robotics","computer science"],"dc:title":["Adaptive Visual Slam for Indoor Environments"],"dc:type":["Text","Dissertation","Video"]},"updated_at":"2026-08-21T22:21:56Z"}