[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-86078-en":3,"doc-seo-86078-105":29,"detail-sidebar-cat-0-en-105":90},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":11,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},86078,2336464648746,"Skyler","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","Mapping Pamir: Multi-Session Visual-Inertial SLAM and 3D Reconstruction of an Underwater Shipwreck","A framework enables multi-session mapping of underwater environments using an affordable action camera paired with water-depth measurements from a dive computer. Visual-inertial data are processed with the open-source SVIn2 VI-SLAM system to produce per-session trajectories and sparse reconstructions. Keyframes and estimated poses feed a COLMAP SfM pipeline for global optimization and dense reconstruction. When calibration targets are available, coordinate transforms align sessions to a shared reference frame, demonstrated on the Pamir shipwreck near Barbados.","Mapping Pamir: Multi-Session Visual-Inertial SLAM and 3D Reconstruction of an Underwater Shipwreck  \nMichalis Chatzispyroua∗, Luke Horganb∗, Hyunkil Hwanga∗, Harish Sathishchandrab∗, Chinmay Burgula , Monika Roznerec , Alberto Quattrini Lid , Philippos Mordohaib , Ioannis Rekleitisa  \narXiv :2607 . 10925v1 [ cs .RO] 12 Jul 2026  \nAbstract—This paper presents a framework for multi-session mapping of underwater environments utilizing an affordable action camera. The Visual-Inertial data are augmented by water depth recordings from a dive computer. SVIn2, an opensource VI-SLAM framework is utilized to generate a trajectory and a sparse reconstruction for each session. Utilizing the keyframes extracted from SVIn2, and the estimated camera poses, a Structure-from-Motion (SfM) framework – COLMAP – is employed for global optimization and produce a dense reconstruction of the target environment. The presence of calibration targets at fixed locations, when available, is used to estimate the coordinate transformation between different data collection sessions, thus transforming the different sessions into the same coordinate frame. The proposed pipeline is employed for the mapping of a shipwreck off the coast of Barbados. For the first time, both the exterior and the accessible interior parts of the wreck were mapped in two sessions, while a third session employed two cameras with different fields of view.  \nI. INTRODUCTION  \nAccurate mapping of underwater structures is crucial for several domains, including underwater archaeology [1], [2],[3], [4], off-shore energy platform inspection [5], and environmental monitoring [6] . However, underwater vision has proven to be extremely challenging [7], [8],[9] . Additionally, the deployment of an autonomous underwater vehicle (AUV) is expensive and time consuming. Early work proposed the use of an inexpensive action camera [10] to deploy SVIn2 [11], a robust Visual-Inertial (VI) SLAM framework based on OKVIS [12] . The advantage of the specific camera (GoPro Hero9-Hero13) is that it encodes video at 30 fps (or higher) and Inertial Measurement Unit (IMU) data at 100 Hz in a single video file, thereby allowing VIO and VI-SLAM algorithms to be run on it. Please refer to the work of Joshi et al. [10] for more details and a comparison of different opensource VIO/VI-SLAM packages. From the formulation of the VI-SLAM problem, roll and pitch are observable, but the position and yaw orientation are not [13], [14] . Although in SVIn2 [11] a water-depth sensor was used, the GoPro action camera does not have one. However, since most divers carry  \n∗The first four authors have contributed equally to this work and are listed in alphabetical order. a University of Delaware, Newark, DE, USA, {michalis,hkhwang,cmburgul,[yiannisr](yiannisr}@udel.edu)[}](yiannisr}@udel.edu)[@udel.edu](yiannisr}@udel.edu) . b Stevens Institute of Technology, Hoboken, NJ, USA,  \n{lhorgan,hsathish,[pmordoha](pmordoha}@stevens.edu)[}](pmordoha}@stevens.edu)[@stevens.edu](pmordoha}@stevens.edu)  \nc Binghamton University, Binghamton, NY, USA mrozner1@binghamton .edu d Dartmouth College, Hanover, NH, [USA](USA alberto.quattrini.li@dartmouth.edu)[ alberto.quattrini.li@dartmouth.edu](USA alberto.quattrini.li@dartmouth.edu)  \nThis research has been supported in part by the National Science Foundation under grants 1943205, 2024541, 2024653, and 2024741 . The authors are also grateful for equipment support by Halcyon Dive Systems, Teledyne FLIR LLC, and KELDAN GmbH lights.  \nFig. 1: GoPro setup deployed over the Pamir shipwreck, Barbados.  \na dive computer that records the depth in the water during the dive, this information is available for each deployment.  \nThe challenging nature of underwater environments makes extensive deployments quite difficult; nitrogen loading and the danger of decompression sickness in conjunction with the amount of breathing gas a diver can carry limit the available time underwater. This requires breaking the dat","cbCailoIelqrs9op","https://ap.wps.com/l/cbCailoIelqrs9op","pdf",5339999,4,1,"English","en",105,"# Introduction\n## Motivation and challenges in underwater mapping\n## Action-camera based VI-SLAM and depth correction\n## Multi-dive processing and dense reconstruction pipeline\n## Contributions and experimental application","[{\"question\":\"How does the pipeline obtain absolute underwater depth information for SLAM?\",\"answer\":\"SVIn2 uses recordings from video inertial sensors, but it lacks direct depth sensing. The method corrects the Z-axis (depth) estimates using the water-depth values recorded by a dive computer during each deployment.\"},{\"question\":\"What role do SVIn2 keyframes play in producing dense 3D reconstructions?\",\"answer\":\"SVIn2 generates trajectories and sparse reconstructions per session. The extracted keyframes, along with estimated camera poses, are then used as inputs to COLMAP to perform global optimization and generate dense reconstructions.\"},{\"question\":\"How are multiple data collection sessions aligned into a common coordinate frame?\",\"answer\":\"When fixed calibration targets are available across sessions, coordinate transformations between camera measurements and those targets are used to convert different sessions into the same coordinate frame.\"}]",1784208372,20,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"mapping-pamir-multi-session-visual-inertial-slam-and-3d-reconstruction-of-an-underwater-shipwreck","",{"@graph":35,"@context":84},[36,52,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":20},"https://docshare.wps.com/document/mapping-pamir-multi-session-visual-inertial-slam-and-3d-reconstruction-of-an-underwater-shipwreck/86078/",{"url":51,"name":13,"@type":53,"author":54,"headline":13,"publisher":56,"fileFormat":59,"inLanguage":23,"description":14,"dateModified":60,"datePublished":61,"encodingFormat":59,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":55},"Person",{"url":40,"name":57,"@type":58},"DocShare","Organization","application/pdf","2026-07-27","2026-07-16",true,{"@type":64,"interactionType":65,"userInteractionCount":20},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"How does the pipeline obtain absolute underwater depth information for SLAM?","Question",{"text":74,"@type":75},"SVIn2 uses recordings from video inertial sensors, but it lacks direct depth sensing. The method corrects the Z-axis (depth) estimates using the water-depth values recorded by a dive computer during each deployment.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"What role do SVIn2 keyframes play in producing dense 3D reconstructions?",{"text":79,"@type":75},"SVIn2 generates trajectories and sparse reconstructions per session. The extracted keyframes, along with estimated camera poses, are then used as inputs to COLMAP to perform global optimization and generate dense reconstructions.",{"name":81,"@type":72,"acceptedAnswer":82},"How are multiple data collection sessions aligned into a common coordinate frame?",{"text":83,"@type":75},"When fixed calibration targets are available across sessions, coordinate transformations between camera measurements and those targets are used to convert different sessions into the same coordinate frame.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":57,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,119,122,126,129,133],{"id":21,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":20,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":28,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":28,"doc_module":4,"doc_module_name":45,"category_name":127,"show_sort_weight":28,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":45,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":45,"category_name":135,"show_sort_weight":105,"slug":136},19,"General","general"]