[{"data":1,"prerenderedAt":312},["ShallowReactive",2],{"method-vegatorres2024slam2ref":3},{"method":4,"reference":66,"equipment":86,"figures":117,"results":158},{"id":5,"label":6,"shortName":7,"title":8,"year":9,"era":10,"cluster":11,"scope":12,"keyIdeaZh":13,"keyIdeaEn":14,"fulltextStatus":15,"publicationStatus":16,"recommendation":17,"constructionRelevance":18,"validationEnvironment":19,"strengths":22,"limitations":26,"sensors":37,"platform":40,"estimator":42,"association":43,"timeModel":44,"deskew":45,"loopClosure":46,"globalOptimization":47,"mapRepresentation":48,"prior":49,"outputGeometry":50,"compute":51,"codeUrl":52,"codeLicense":53,"relatedVersions":54},"vegatorres2024slam2ref","Vega-Torres et al., 2024","SLAM2REF","SLAM2REF: advancing long-term mapping with 3D LiDAR and reference map integration for precise 6-DoF trajectory estimation and map extension",2024,"recent","C11b","localization_in_prior_map_or_bim","SLAM2REF 把行動 LiDAR 與 IMU 資料和既有 BIM 或點雲參考圖整合，用於室內無 GPS 環境的長期建圖。流程先由參考圖產生佔據網格與模擬 LiDAR 掃描作為「參考工作段」，再以 DLIO 去除實測掃描的運動畸變，接著用室內版 Scan Context 描述子與 YawGICP 找跨工作段對應，透過多工作段錨定（multi-session anchoring）位姿圖最佳化把漂移的 SLAM 結果對齊參考圖，最後逐幀以點對點 ICP 對齊 1 cm 密度的參考點雲。對齊後以 OctoMap 分析新增與移除的構件並網格化，並允許地圖延伸到參考圖範圍之外。","SLAM2REF aligns drifted LiDAR-inertial sessions to a BIM or TLS reference map through simulated reference sessions, Indoor Scan Context, YawGICP, multi-session anchoring and a final dense ICP, then detects positive and negative changes and extends the map.","full_text_reviewed","peer_reviewed_published","main_body","以 ConSLAM 施工中建物資料集（手持設備、四個序列）驗證，並以 seq 2 的 TLS 建立約半公分精度的 BIM（as-built，非設計模型）。延續 [bimslam2023] 的 BIM-SLAM 路線；作者說明相較 BIM-SLAM 增加大型參考圖、IMU 去畸變、最終 ICP 與容許掃描-地圖差異的能力（Sec. 7）。",[20,21],"real_construction_site","public_benchmark",[23,24,25],"Automatic alignment without manual initialization of the first keyframe, unlike ground-truth generation in ConSLAM or Newer College (Sec. 7)","Alignment to a clutter-free BIM despite scan-map deviations from clutter, furniture and dynamic objects (Sec. 1, 6)","Not restricted to Manhattan-world layouts; supports extending the reference map (Sec. 7)",[27,28,29,30,31,32,33,34,35,36],"Sensitive to the initial SLAM or LIO poses; large drift, especially Z-drift in narrow corridors without floor or ceiling points, may not be corrected (Sec. 8, Fig. 15)","Large deviations of permanent walls or columns, low overlap or symmetric environments can defeat alignment; the final ICP may be wrong where the reference map is wrong (Sec. 8)","Not real-time; final ICP can take several dozen minutes (Sec. 8)","Indoor Scan Context needs a 360-degree horizontal FoV, so solid-state LiDARs and depth cameras are not directly supported (Sec. 8)","Window reflections create fictitious elements in change detection (Sec. 8)","Only clutter and dynamic-object deviations are addressed, not alterations of permanent building elements (Sec. 1)","After the KNN loops, rotational APE rises in S3 and S5 (Table 1, Sec. 6); Sec. 7 explains this pattern (naming it for sequences 2 and 5 there) by erroneous KNN loops detected where the ISC-aligned trajectory, before Umeyama alignment, still deviated about 1.5 m in Z and X from the ground truth; the final ICP filters these loops (Sec. 6, Sec. 7, Table 1)","Correct ISC correspondences are very sensitive to the number of top candidates (N_c = 100) (Sec. 5.2.2)","(inference) Final-ICP poses to the TLS map serve as ground truth, so the TLS-referenced final stage is not independently evaluated","(inference) The ConSLAM BIM was modelled from the TLS point cloud of sequence 2 (Sec. 5.1), i.e. an as-built model, so design-versus-as-built discrepancy (brief Sec. 11 risk 2) is not tested; the same caveat is recorded for stuhrenberg2025liobim in C11b",[38,39],"3D LiDAR of the ConSLAM handheld system (model not reported; the ISC descriptor requires a 360-degree horizontal FoV)","9-axis IMU of the ConSLAM handheld system (model not reported; used for DLIO deskewing; LiDAR-IMU extrinsics from OA-LICalib)",[41],"handheld (ConSLAM sequences)","DLIO front end for deskewing and odometry; multi-session anchoring pose graph in GTSAM (iSAM2, batch) with odometry, Indoor Scan Context and KNN inter-session constraints; final point-to-point ICP of each scan to a 1 cm dense reference cloud (Sec. 4.2, 5.2)","Indoor Scan Context descriptors (binary occupancy per bin, 60 sectors x 20 rings, at least 40 points, 10 m radius) matched against simulated reference scans: 100 top candidates from a nanoflann KD-tree of rotation-invariant 1D descriptors, column-wise cosine score with threshold 0.3 and yaw shifts limited to 36 deg, then YawGICP (built on Open3D GICP) against 3-scan reference submaps; KNN submap loops (K = 5) with adaptive covariance, omitted for BIM references; final point-to-point ICP to a 1 cm dense reference cloud with fitness at 1 cm and 3 cm computed on points within 30 cm (Sec. 4.2.2, 4.2.3, 5.2.2)","continuous-time deskew inherited from DLIO (constant jerk and angular acceleration with IMU); discrete keyframe poses in the pose graph (Sec. 4.2.1.1)","IMU-based point-wise motion correction using DLIO; bags replayed at half speed to avoid deskew errors (Sec. 4.2.1.1, 5.2.2.1)","inter-session loops to reference-map sessions (ISC, then KNN); intra-session loops optional from the SLAM front end (Sec. 4.2)","multi-session anchoring pose-graph optimization, then per-scan final ICP to the dense reference cloud (Sec. 4.2.3)","point cloud; OctoMap for dynamic-object removal and free-space reasoning; voxel-cube meshes for positive and negative differences (Sec. 4.3)","BIM (IFC, filtered to permanent elements such as walls, columns, slabs and floors; doors and windows excluded) or TLS point cloud as reference map; occupancy grid from IfcConvert SVG sections, and Blensor-simulated 360-deg scans at skeleton-sampled locations (vertical FoV -45 to 45 deg for TLS references, 0 to -25 deg without ceiling for BIM; noise std 0.03 m, angular resolution 0.1728 deg, max range 15 m) (Sec. 4.1, 5.2.1)","6-DoF poses in the reference-map frame with per-scan alignment classes (perfect, good, bad, outside map); aligned and extended point cloud map; meshes of positive and negative differences (Sec. 4.2.3, 4.3)","offline, not real-time; the final ICP stage can take several dozen minutes (Sec. 8); Step 2 in C++ with OpenMP (parallel YawGICP and final ICP), Steps 1 and 3 in Python (Trimesh, OctoMap, Open3D); GTSAM iSAM2; compute hardware not reported (Sec. 5.2)","https:\u002F\u002Fgithub.com\u002FMigVega\u002FSLAM2REF","GPL-3.0",[55,59,62],{"relation":56,"title":57,"doi_or_url":58},"preprint","arXiv 2408.15948 v1 (posted 2024-08-28, after journal publication; journal_ref Construction Robotics 2024)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2408.15948",{"relation":60,"title":61,"doi_or_url":52},"code_release","SLAM2REF repository (GPL-3.0)",{"relation":63,"title":64,"doi_or_url":65},"dataset","ConSLAM BIM and GT Poses (dataset, TUM mediaTUM, 2024)","10.14459\u002F2024mp1743877",{"id":5,"kind":67,"shortName":7,"title":8,"authors":68,"year":9,"venue":72,"venueType":73,"publisher":74,"volumeIssuePages":75,"doi":76,"arxivId":77,"url":58,"firstPublicDate":78,"publicationStatus":16,"metadataStatus":79,"fulltextStatus":15,"era":10,"classicReason":80,"codeUrl":52,"cluster":11,"topics":81,"mdpi":82,"verification":83,"label":6,"fulltextRoute":84,"versionRead":85,"addedByCensus":82},"method",[69,70,71],"Miguel A. Vega-Torres","Alexander Braun","André Borrmann","Construction Robotics","journal","Springer","8(2), article 13","10.1007\u002Fs41693-024-00126-w","2408.15948","2024-07-05","metadata_verified","not_applicable",[11],false,"corrected","arXiv","arXiv 2408.15948v1 HTML (posted 2024-08-28, after journal publication), read in full; Springer version of record (Construction Robotics 8, article 13, published 2024-07-05, CC BY 4.0) HTML compared for Table 1, Sec. 5.1, Sec. 6 (14.8 cm and 0.56 deg) and Sec. 8 passages, which match",[87,94,99,104,106,112],{"category":88,"model":89,"canonical":89,"role":90,"dataset":91,"specs":92,"locator":93},"lidar","not_reported (3D LiDAR of the ConSLAM handheld system)","method input","ConSLAM","scans deskewed with DLIO; rosbags replayed at half speed; method assumes 360-deg horizontal FoV","Sec. 5.1, 5.2.2, 8",{"category":95,"model":96,"canonical":96,"role":90,"dataset":91,"specs":97,"locator":98},"imu","not_reported (9-axis IMU of the ConSLAM handheld system)","9-axis; LiDAR-IMU extrinsics estimated with OA-LICalib","Sec. 5.1",{"category":100,"model":101,"canonical":101,"role":102,"dataset":91,"specs":103,"locator":98},"camera","not_reported (RGB camera of the ConSLAM handheld system)","dataset sensor","not_reported; not used by SLAM2REF",{"category":100,"model":105,"canonical":105,"role":102,"dataset":91,"specs":103,"locator":98},"not_reported (near-infrared camera of the ConSLAM handheld system)",{"category":107,"model":108,"canonical":108,"role":109,"dataset":91,"specs":110,"locator":111},"tls_scanner","not_reported (TLS point clouds supplied with ConSLAM)","reference or ground truth","per-sequence TLS clouds used as reference maps; S2 TLS cloud used to model a half-centimetre-accurate BIM","Sec. 1, 5.1, 6",{"category":113,"model":114,"canonical":114,"role":102,"dataset":91,"specs":115,"locator":116},"platform","handheld system (model not reported)","four construction-site sequences S2-S5 of 225-340 m","Sec. 5.1, Table 1",[118,131,140,150],{"refId":5,"refLabel":6,"fig":119,"whatZh":120,"license":121,"licenseUrl":122,"sourceUrl":123,"src":124,"width":125,"height":126,"thumb":127,"thumbWidth":128,"thumbHeight":129,"modified":130},"Fig. 1","SLAM2REF 三步驟流程總覽：由參考圖產生工作段資料、參考圖多工作段錨定、變化偵測與地圖更新。","CC BY 4.0","https:\u002F\u002Fcreativecommons.org\u002Flicenses\u002Fby\u002F4.0\u002F","https:\u002F\u002Fmedia.springernature.com\u002Flw1200\u002Fspringer-static\u002Fimage\u002Fart%3A10.1007%2Fs41693-024-00126-w\u002FMediaObjects\u002F41693_2024_126_Fig1_HTML.png","\u002Ffigure-files\u002Fvegatorres2024slam2ref\u002Ffig-1.webp",1200,456,"\u002Ffigure-files\u002Fvegatorres2024slam2ref\u002Ffig-1.thumb.webp",480,182,"converted to WebP",{"refId":5,"refLabel":6,"fig":132,"whatZh":133,"license":121,"licenseUrl":122,"sourceUrl":134,"src":135,"width":136,"height":137,"thumb":138,"thumbWidth":128,"thumbHeight":139,"modified":130},"Fig. 4","由 BIM 網格模擬的 LiDAR 掃描與對應的 Scan Context 描述子，構成合成的參考工作段。","https:\u002F\u002Farxiv.org\u002Fhtml\u002F2408.15948v1\u002FS1_3.png","\u002Ffigure-files\u002Fvegatorres2024slam2ref\u002Ffig-4.webp",525,320,"\u002Ffigure-files\u002Fvegatorres2024slam2ref\u002Ffig-4.thumb.webp",293,{"refId":5,"refLabel":6,"fig":141,"whatZh":142,"license":121,"licenseUrl":122,"sourceUrl":143,"src":144,"width":145,"height":146,"thumb":147,"thumbWidth":128,"thumbHeight":148,"modified":149},"Fig. 13","ConSLAM 序列 2 至 5 對齊 TLS 點雲，以及序列 2 對齊 BIM 後的軌跡、點雲地圖與新增元素。","https:\u002F\u002Farxiv.org\u002Fhtml\u002F2408.15948v1\u002FresultedMaps6_compressed.png","\u002Ffigure-files\u002Fvegatorres2024slam2ref\u002Ffig-13.webp",1400,1044,"\u002Ffigure-files\u002Fvegatorres2024slam2ref\u002Ffig-13.thumb.webp",358,"resized to at most 1400 px wide and converted to WebP",{"refId":5,"refLabel":6,"fig":151,"whatZh":152,"license":121,"licenseUrl":122,"sourceUrl":153,"src":154,"width":125,"height":155,"thumb":156,"thumbWidth":128,"thumbHeight":157,"modified":130},"Fig. 14","序列 2 對齊後的變化偵測結果：圖 a 與 TLS 點雲比較，圖 b 與 BIM 比較，新增元素以紅色標示。","https:\u002F\u002Fmedia.springernature.com\u002Flw1200\u002Fspringer-static\u002Fimage\u002Fart%3A10.1007%2Fs41693-024-00126-w\u002FMediaObjects\u002F41693_2024_126_Fig14_HTML.jpg","\u002Ffigure-files\u002Fvegatorres2024slam2ref\u002Ffig-14.webp",374,"\u002Ffigure-files\u002Fvegatorres2024slam2ref\u002Ffig-14.thumb.webp",150,{"totalRows":159,"groupCount":160,"groups":161,"others":311},22,2,[162,286],{"slug":163,"group":164,"sourceId":5,"sourceLabel":6,"table":165,"selfRows":166,"metrics":167,"seqs":176,"entrants":188,"cells":200,"outcomes":280,"locators":281,"hardware":282,"wordings":283,"notes":284},"vegatorres2024slam2ref-table-1","vegatorres2024slam2ref:Table 1","Table 1",20,[168,173],{"label":169,"unit":170,"statistic":171,"alignment":172},"translational APE RMSE","cm","RMSE","not_reported",{"label":174,"unit":175,"statistic":171,"alignment":172},"rotational (angular) APE RMSE","deg",[177,180,182,184,186],{"dataset":91,"sequence":178,"environment":179},"S2 (225 m)","construction site, building under construction (indoor, handheld)",{"dataset":91,"sequence":181,"environment":179},"S3 (340 m)",{"dataset":91,"sequence":183,"environment":179},"S4 (275 m)",{"dataset":91,"sequence":185,"environment":179},"S5 (320 m)",{"dataset":91,"sequence":187,"environment":179},"Average",[189,193,195,197],{"name":190,"methodId":191,"linkable":192,"proposed":82,"self":82},"DLIO","dlio2023",true,{"name":194,"methodId":5,"linkable":192,"proposed":192,"self":192},"SC (DLIO after Indoor Scan Context loop detection and optimization; SLAM2REF intermediate stage)",{"name":196,"methodId":5,"linkable":192,"proposed":192,"self":192},"KNN (after KNN loops and optimization; SLAM2REF intermediate stage)",{"name":198,"methodId":199,"linkable":82,"proposed":82,"self":82},"ConSLAM (ground-truth poses supplied with the dataset)","trzeciak2023conslam",[201,205,208,210,212,214,216,219,221,224,226,228,229,231,232,234,235,237,238,240,241,243,245,247,249,251,253,255,257,259,261,263,265,267,268,270,272,274,276,278],[202,202,202,203,204,202,204,204,202],0,20.2,-1,[202,206,202,207,204,202,204,204,202],1,2.2,[202,202,206,209,204,202,204,204,202],21.4,[202,206,206,211,204,202,204,204,202],2.6,[202,202,160,213,204,202,204,204,202],359,[202,206,160,215,204,202,204,204,202],6.4,[202,202,217,218,204,202,204,204,202],3,17.4,[202,206,217,220,204,202,204,204,202],2.3,[202,202,222,223,204,202,204,204,202],4,104.5,[202,206,222,225,204,202,204,204,202],3.4,[206,202,202,227,204,202,204,204,202],20.1,[206,206,202,207,204,202,204,204,202],[206,202,206,230,204,202,204,204,202],24.3,[206,206,206,211,204,202,204,204,202],[206,202,160,233,204,202,204,204,202],358.6,[206,206,160,215,204,202,204,204,202],[206,202,217,236,204,202,204,204,202],18.7,[206,206,217,220,204,202,204,204,202],[206,202,222,239,204,202,204,204,202],105.4,[206,206,222,225,204,202,204,204,202],[160,202,202,242,204,202,204,204,202],9,[160,206,202,244,204,202,204,204,202],1.4,[160,202,206,246,204,202,204,204,202],34.6,[160,206,206,248,204,202,204,204,202],4.3,[160,202,160,250,204,202,204,204,202],53.4,[160,206,160,252,204,202,204,204,202],3.5,[160,202,217,254,204,202,204,204,202],11.2,[160,206,217,256,204,202,204,204,202],2.5,[160,202,222,258,204,202,204,204,202],27.1,[160,206,222,260,204,202,204,204,202],2.9,[217,202,202,262,204,202,204,204,202],5.2,[217,206,202,264,204,202,204,204,202],0.7,[217,202,206,266,204,202,204,204,202],4.2,[217,206,206,264,204,202,204,204,202],[217,202,160,269,204,202,204,204,202],9.3,[217,206,160,271,204,202,204,204,202],0.9,[217,202,217,273,204,202,204,204,202],12.1,[217,206,217,275,204,202,204,204,202],1.1,[217,202,222,277,204,202,204,204,202],7.7,[217,206,222,279,204,202,204,204,202],0.8,[],[165],[],[],[285],"ConSLAM S2-S5 aligned to the per-sequence TLS point cloud; APE RMSE against the SLAM2REF final-ICP poses, which the authors use as ground truth; evo, TUM format, Umeyama alignment (scale handling not stated)",{"slug":287,"group":288,"sourceId":5,"sourceLabel":6,"table":289,"selfRows":160,"metrics":290,"seqs":294,"entrants":296,"cells":299,"outcomes":304,"locators":305,"hardware":307,"wordings":308,"notes":309},"vegatorres2024slam2ref-text-sec-6","vegatorres2024slam2ref:Text Sec.6","Text Sec.6",[291,292],{"label":169,"unit":170,"statistic":171,"alignment":172},{"label":293,"unit":175,"statistic":171,"alignment":172},"rotational APE RMSE",[295],{"dataset":91,"sequence":178,"environment":179},[297],{"name":298,"methodId":5,"linkable":192,"proposed":192,"self":192},"SLAM2REF with BIM reference (after final ICP)",[300,302],[202,202,202,301,204,202,204,204,202],14.8,[202,206,202,303,204,202,204,204,202],0.56,[],[306],"Sec. 6, Fig. 12",[],[],[310],"ConSLAM S2 aligned to a BIM modelled from the S2 TLS cloud; APE RMSE after the final ICP against the TLS-derived reference poses",[],1790510654980]