[{"data":1,"prerenderedAt":515},["ShallowReactive",2],{"method-lamm2025":3},{"method":4,"reference":57,"equipment":86,"figures":122,"results":123},{"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":21,"limitations":27,"sensors":35,"platform":38,"estimator":41,"association":42,"timeModel":43,"deskew":44,"loopClosure":45,"globalOptimization":46,"mapRepresentation":47,"prior":48,"outputGeometry":49,"compute":50,"codeUrl":51,"codeLicense":52,"relatedVersions":53},"lamm2025","Wei et al., 2025b","LAMM","Large-Scale Multi-Session Point-Cloud Map Merging",2025,"recent","C06","offline_map_refinement","LAMM 是離線的多時段光達點雲地圖合併框架，輸入各代理人由前端 SLAM（如 FAST-LIO2）得到的掃描與初始位姿。先以 M-Detector 為基礎，在正向與反向時間序列各做一次遮擋測試的雙向濾波移除動態點；再以 BTC 描述子在各序列的資料庫中搜尋序列內與序列間迴圈；序列間迴圈以「用每個迴圈把另一序列起點投影到本序列座標」後做 RANSAC 聚類來剔除離群；最後依連通性把序列分組，對每組子位姿圖以 GTSAM 做位姿圖最佳化，輸出一或多張全域一致的點雲地圖。","LAMM merges multi-agent, multi-session LiDAR sub-maps from different LiDAR types, adding temporal bidirectional dynamic filtering and robust loop outlier removal.","full_text_reviewed","peer_reviewed_published","background","未於工地測試。資料為 KITTI 市區、HeLiPR 住宅區（Town、Roundabout）、WildPlaces 自然環境，以及深圳北站附近以背包式設備（Hesai 128 線光達與四部 Hikvision 相機）蒐集的都市彩色點雲（Sec. IV、Table I）。能把機械旋轉式與 Livox Avia 等不同掃描型態、不同時間的序列合併成一張地圖，符合工地多次、多設備掃描整合的需求（推論）；但只以位姿圖合併，重疊區的局部幾何一致性沒有以參考點雲量化。",[20],"public_benchmark",[22,23,24,25,26],"Merged all six KITTI test sequences; BTC-only merging failed on 00 and 02 and Disco-SLAM and DCL-SLAM failed on 02 (Sec. IV-A1, Table II)","ATE RMSE 2.040 m on KITTI 00 and 4.746 m on 02 versus 3.330 m and 8.749 m for a single FAST-LIO2 run (Table II)","Bi-M-Detector gave the most balanced static and dynamic classification on 141 HeLiPR Town01 frames (AA 80.65%, HA 79.86%) (Sec. IV-A3, Table IV)","Merged Ouster OS2-128 and Livox Avia sequences into one map on HeLiPR Town and Roundabout (merged multi-LiDAR ATE 4.008 m and 2.101 m) (Table V)","Open-source code (Sec. I)",[28,29,30,31,32,33,34],"Merging uses pose graph optimization only, with no map-level refinement of overlaps (Sec. III-E3); follow-up work reports local divergence in overlaps (lemon2026 Sec. II-B, Sec. VII-C, Fig. 9)","Bi-M-Detector is not best in static accuracy (92.86%) or dynamic accuracy (70.05%) alone (Table IV)","Without the loop filter, merging fails on KITTI 00 and 02 (Sec. IV-B, Table II)","Dynamic-removal evaluation uses 141 manually selected frames of one HeLiPR sequence (Sec. IV-A3)","No open-source map-merging baseline was available; comparison is against multi-robot SLAM systems and BTC-only merging (Sec. IV-A1)","WildPlaces and Shenzhen results and HeLiPR merged maps appear only in supplementary material; efficiency improvement left to future work (Sec. IV-C, V)","Runtime not reported",[36,37],"3D LiDAR of different scanning patterns: Velodyne (KITTI, WildPlaces), Ouster OS2-128 and Livox Avia (HeLiPR), Hesai 128-line (Shenzhen)","four Hikvision cameras on the Shenzhen backpack, used with R3LIVE to produce colored point clouds",[39,40],"backpack device (self-collected Shenzhen dataset)","public datasets KITTI, HeLiPR and WildPlaces (carrier platforms not described in the paper)","Offline back end: bidirectional M-Detector dynamic removal per sequence, BTC place recognition, false-loop filtering, connectivity check that splits sequences into sub-pose graphs, and standard pose graph optimization of each sub-pose graph with the first node anchored to a reference pose, solved in GTSAM (Sec. III-B to III-E, Eq. 2).","BTC (Binary Triangle Combined) descriptors stored in one database per loaded sequence; each scan is searched against all databases in descending order for inner- and inter-sequence loops; rough detection by hash-table matching, fine detection by clustering transforms between triangle pairs, geometric verification by point-to-plane distance using fewer than 50 key points (Sec. III-D).","discrete scan poses from the front-end odometry (Sec. III-A)","not performed by LAMM; inputs are registered scans and poses from a front-end SLAM such as FAST-LIO2 (Sec. III-A, III-B)","Inner-sequence loops are rejected when the initial poses of the matched frames are geographically far apart; for inter-sequence loops, each loop projects the start position of one sequence into the other sequence's frame (Eq. 1) and RANSAC clustering of these projected points removes outliers (Sec. III-E1, Fig. 3).","Pose graph optimization per connected sub-pose graph (odometry and intra- and inter-sequence loop edges, anchor prior on the first node), solved with GTSAM; no bundle adjustment or map-level refinement (Sec. III-E2, III-E3).","registered LiDAR scans with poses per sequence; output is one merged global point-cloud map per connected group of sequences (Sec. III-B)","LiDAR scans with initial poses from a front-end SLAM (FAST-LIO2 for KITTI and HeLiPR; R3LIVE for Shenzhen); no prior relative pose between sequences, which may start anywhere (Sec. III-A, IV-A1, IV-C)","merged multi-session point-cloud map (abstract)","Offline, C++; all experiments on a desktop with an Intel i9-13900K @ 3.0 GHz and 128 GB memory (Sec. IV); runtime is not reported.","https:\u002F\u002Fgithub.com\u002Fhku-mars\u002FLAMM","not_verified",[54],{"relation":55,"title":56,"doi_or_url":51},"code_release","hku-mars\u002FLAMM",{"id":5,"kind":58,"shortName":7,"title":8,"authors":59,"year":9,"venue":71,"venueType":72,"publisher":73,"volumeIssuePages":74,"doi":75,"arxivId":76,"url":77,"firstPublicDate":78,"publicationStatus":16,"metadataStatus":79,"fulltextStatus":15,"era":10,"classicReason":80,"codeUrl":51,"cluster":11,"topics":81,"mdpi":82,"verification":83,"label":6,"fulltextRoute":84,"versionRead":85,"addedByCensus":82},"method",[60,61,62,63,64,65,66,67,68,69,70],"Hairuo Wei","Rundong Li","Yixi Cai","Chongjian Yuan","Yunfan Ren","Zuhao Zou","Huajie Wu","Chunran Zheng","Shunbo Zhou","Kaiwen Xue","Fu Zhang","IEEE Robotics and Automation Letters","journal","IEEE","10(1):88-95","10.1109\u002Flra.2024.3504317",null,"https:\u002F\u002Fdoi.org\u002F10.1109\u002FLRA.2024.3504317","2024-11-21","metadata_verified","not_applicable",[11],false,"corrected","NTU institutional (Chrome)","Version of record, IEEE RA-L 10(1):88-95, published 21 November 2024 (IEEE Xplore HTML full text; Tables I-V read from the IEEE large table images)",[87,93,100,105,109,115,118],{"category":88,"model":89,"canonical":89,"role":90,"dataset":76,"specs":91,"locator":92},"compute","Intel i9-13900K","compute for runtime","3.0 GHz; 128 GB memory; desktop computer","Sec. IV",{"category":94,"model":95,"canonical":95,"role":96,"dataset":97,"specs":98,"locator":99},"platform","backpack device","method input","Shenzhen (self-collected)","carries a Hesai 128-line LiDAR and four Hikvision cameras","Sec. IV-C2",{"category":101,"model":102,"canonical":102,"role":96,"dataset":97,"specs":103,"locator":104},"lidar","Hesai 128-line LiDAR","128 lines","Sec. IV-C2; Table I",{"category":106,"model":107,"canonical":107,"role":96,"dataset":97,"specs":108,"locator":99},"camera","Hikvision cameras (four)","four cameras; colored point cloud via R3LIVE",{"category":101,"model":110,"canonical":110,"role":111,"dataset":112,"specs":113,"locator":114},"Livox Avia","dataset sensor","HeLiPR (Town, Roundabout)","not_reported","Sec. IV-C1; Table I",{"category":101,"model":116,"canonical":117,"role":111,"dataset":112,"specs":113,"locator":114},"OS2-128","Ouster OS2-128",{"category":101,"model":119,"canonical":119,"role":111,"dataset":120,"specs":113,"locator":121},"Velodyne (model not named)","KITTI; WildPlaces","Table I",[],{"totalRows":124,"groupCount":125,"groups":126,"others":505},40,6,[127,217,312,458],{"slug":128,"group":129,"sourceId":130,"sourceLabel":131,"table":132,"selfRows":133,"metrics":134,"seqs":145,"entrants":154,"cells":159,"outcomes":211,"locators":212,"hardware":213,"wordings":214,"notes":215},"lemon2026-table-v","lemon2026:Table V","lemon2026","Wang et al., 2026","Table V",12,[135,138,140,143],{"label":136,"unit":137,"statistic":113,"alignment":113},"AWD(m)","m",{"label":139,"unit":137,"statistic":113,"alignment":113},"CD(m)",{"label":141,"unit":142,"statistic":113,"alignment":113},"SCS","unitless",{"label":144,"unit":142,"statistic":113,"alignment":113},"MME",[146,150,152],{"dataset":147,"sequence":148,"environment":149},"MARS-LVIG","Island","large-scale aerial scene over 100,000 m^2",{"dataset":147,"sequence":151,"environment":149},"Town",{"dataset":147,"sequence":153,"environment":149},"Airport",[155,157],{"name":7,"methodId":5,"linkable":156,"proposed":82,"self":156},true,{"name":158,"methodId":130,"linkable":156,"proposed":156,"self":82},"Ours",[160,164,167,170,173,175,177,179,181,183,185,187,189,191,193,195,197,199,201,203,205,207,208,209],[161,161,161,162,163,161,163,163,161],0,1.3,-1,[161,165,161,166,163,161,163,163,161],1,6.44,[161,168,161,169,163,161,163,163,161],2,0.5,[161,171,161,172,163,161,163,163,161],3,-5.51,[165,161,161,174,163,161,163,163,161],0.3,[165,165,161,176,163,161,163,163,161],0.43,[165,168,161,178,163,161,163,163,161],0.49,[165,171,161,180,163,161,163,163,161],-6.13,[161,161,165,182,163,161,163,163,161],1.77,[161,165,165,184,163,161,163,163,161],10.64,[161,168,165,186,163,161,163,163,161],0.57,[161,171,165,188,163,161,163,163,161],-6.37,[165,161,165,190,163,161,163,163,161],0.65,[165,165,165,192,163,161,163,163,161],1.11,[165,168,165,194,163,161,163,163,161],0.47,[165,171,165,196,163,161,163,163,161],-6.52,[161,161,168,198,163,161,163,163,161],0.75,[161,165,168,200,163,161,163,163,161],36.05,[161,168,168,202,163,161,163,163,161],0.62,[161,171,168,204,163,161,163,163,161],-6.17,[165,161,168,206,163,161,163,163,161],0.25,[165,165,168,194,163,161,163,163,161],[165,168,168,186,163,161,163,163,161],[165,171,168,210,163,161,163,163,161],-6.28,[],[132],[],[],[216],"Mapping quality against the MARS-LVIG ground-truth map (DJI L1 LiDAR processed with DJI Terra); AWD average Wasserstein distance, CD Chamfer distance, SCS spatial consistency score, MME mean map entropy; all lower is better; metrics as defined in MapEval",{"slug":218,"group":219,"sourceId":130,"sourceLabel":131,"table":220,"selfRows":221,"metrics":222,"seqs":226,"entrants":247,"cells":252,"outcomes":305,"locators":307,"hardware":308,"wordings":309,"notes":310},"lemon2026-table-iv","lemon2026:Table IV","Table IV",10,[223],{"label":224,"unit":137,"statistic":225,"alignment":113},"RMSE of the ATE (m)","RMSE",[227,231,233,235,237,240,242,244,245,246],{"dataset":228,"sequence":229,"environment":230},"S3E","Campus 3","multi-robot, see Table II of the paper",{"dataset":228,"sequence":232,"environment":230},"Dormitory",{"dataset":228,"sequence":234,"environment":230},"Library",{"dataset":228,"sequence":236,"environment":230},"Tunnel",{"dataset":238,"sequence":239,"environment":230},"GEODE","Inlandwaterways",{"dataset":238,"sequence":241,"environment":230},"Offroad",{"dataset":238,"sequence":243,"environment":230},"Tunnelingtunnel",{"dataset":147,"sequence":153,"environment":230},{"dataset":147,"sequence":151,"environment":230},{"dataset":147,"sequence":148,"environment":230},[248,250,251],{"name":249,"methodId":76,"linkable":82,"proposed":82,"self":82},"DCL-SLAM",{"name":7,"methodId":5,"linkable":156,"proposed":82,"self":156},{"name":158,"methodId":130,"linkable":156,"proposed":156,"self":82},[253,255,257,259,261,263,265,267,269,271,273,274,276,278,279,281,283,285,286,287,288,290,292,294,295,297,298,300,302,303],[161,161,161,254,163,161,163,163,161],17.89,[165,161,161,256,163,161,163,163,161],12.51,[168,161,161,258,163,161,163,163,161],3.51,[161,161,165,260,163,161,163,163,161],3.52,[165,161,165,262,163,161,163,163,161],28.67,[168,161,165,264,163,161,163,163,161],3.46,[161,161,168,266,163,161,163,163,161],6.37,[165,161,168,268,163,161,163,163,161],4.58,[168,161,168,270,163,161,163,163,161],1.48,[161,161,171,272,163,161,163,163,161],3.09,[165,161,171,76,161,161,163,163,161],[168,161,171,275,163,161,163,163,161],0.98,[161,161,277,76,161,161,163,163,161],4,[165,161,277,76,161,161,163,163,161],[168,161,277,280,163,161,163,163,161],7.44,[161,161,282,76,161,161,163,163,161],5,[165,161,282,284,163,161,163,163,161],25.45,[168,161,282,162,163,161,163,163,161],[161,161,125,76,161,161,163,163,161],[165,161,125,202,163,161,163,163,161],[168,161,125,289,163,161,163,163,161],0.16,[161,161,291,76,161,161,163,163,161],7,[165,161,291,293,163,161,163,163,161],8.41,[168,161,291,162,163,161,163,163,161],[161,161,296,76,161,161,163,163,161],8,[165,161,296,76,161,161,163,163,161],[168,161,296,299,163,161,163,163,161],1.49,[161,161,301,76,161,161,163,163,161],9,[165,161,301,76,161,161,163,163,161],[168,161,301,304,163,161,163,163,161],0.801,[306],"failed (RMSE above 30 m)",[220],[],[],[311],"Multi-robot localization; RMSE of ATE (m); failure (x) = any sequence with RMSE above 30 m; GEODE, MARS-LVIG and S3E sequences split into sessions with unknown relative transforms",{"slug":313,"group":314,"sourceId":5,"sourceLabel":6,"table":315,"selfRows":125,"metrics":316,"seqs":318,"entrants":333,"cells":352,"outcomes":451,"locators":453,"hardware":454,"wordings":455,"notes":456},"lamm2025-table-ii","lamm2025:Table II","Table II",[317],{"label":224,"unit":137,"statistic":225,"alignment":113},[319,323,325,327,329,331],{"dataset":320,"sequence":321,"environment":322},"KITTI","00","urban driving",{"dataset":320,"sequence":324,"environment":322},"02",{"dataset":320,"sequence":326,"environment":322},"05",{"dataset":320,"sequence":328,"environment":322},"06",{"dataset":320,"sequence":330,"environment":322},"07",{"dataset":320,"sequence":332,"environment":322},"08",[334,337,339,340,342,344,346,348,350],{"name":335,"methodId":336,"linkable":156,"proposed":82,"self":82},"FAST-LIO2 (single run)","fastlio2_2022",{"name":338,"methodId":76,"linkable":82,"proposed":82,"self":82},"Disco-SLAM",{"name":249,"methodId":76,"linkable":82,"proposed":82,"self":82},{"name":341,"methodId":76,"linkable":82,"proposed":82,"self":82},"BTC (loop-detection-only merging)",{"name":343,"methodId":5,"linkable":156,"proposed":156,"self":156},"LAMM full model",{"name":345,"methodId":76,"linkable":82,"proposed":82,"self":82},"LAMM wo M-detector",{"name":347,"methodId":76,"linkable":82,"proposed":82,"self":82},"LAMM wo loop filter",{"name":349,"methodId":76,"linkable":82,"proposed":82,"self":82},"LAMM-SOLiD",{"name":351,"methodId":76,"linkable":82,"proposed":82,"self":82},"LAMM-PCM",[353,355,357,359,360,362,364,365,367,369,371,372,373,374,376,378,379,381,383,385,387,389,391,393,395,396,398,400,402,404,406,408,410,412,413,415,417,419,421,423,425,427,429,430,432,434,436,437,439,441,443,445,447,449],[161,161,161,354,163,161,163,163,161],3.33,[165,161,161,356,163,161,163,163,161],2.097,[168,161,161,358,163,161,163,163,161],4.024,[171,161,161,76,161,161,163,163,161],[277,161,161,361,163,161,163,163,161],2.04,[282,161,161,363,163,161,163,163,161],2.392,[125,161,161,76,161,161,163,163,161],[291,161,161,366,163,161,163,163,161],2.061,[296,161,161,368,163,161,163,163,161],2.707,[161,161,165,370,163,161,163,163,161],8.749,[165,161,165,76,161,161,163,163,161],[168,161,165,76,161,161,163,163,161],[171,161,165,76,161,161,163,163,161],[277,161,165,375,163,161,163,163,161],4.746,[282,161,165,377,163,161,163,163,161],6.723,[125,161,165,76,161,161,163,163,161],[291,161,165,380,163,161,163,163,161],5.931,[296,161,165,382,163,161,163,163,161],5.748,[161,161,168,384,163,161,163,163,161],1.765,[165,161,168,386,163,161,163,163,161],1.653,[168,161,168,388,163,161,163,163,161],2.783,[171,161,168,390,163,161,163,163,161],1.875,[277,161,168,392,163,161,163,163,161],1.639,[282,161,168,394,163,161,163,163,161],1.788,[125,161,168,392,163,161,163,163,161],[291,161,168,397,163,161,163,163,161],1.658,[296,161,168,399,163,161,163,163,161],1.59,[161,161,171,401,163,161,163,163,161],0.938,[165,161,171,403,163,161,163,163,161],0.973,[168,161,171,405,163,161,163,163,161],0.984,[171,161,171,407,163,161,163,163,161],1.051,[277,161,171,409,163,161,163,163,161],0.941,[282,161,171,411,163,161,163,163,161],0.948,[125,161,171,409,163,161,163,163,161],[291,161,171,414,163,161,163,163,161],0.939,[296,161,171,416,163,161,163,163,161],0.983,[161,161,277,418,163,161,163,163,161],0.956,[165,161,277,420,163,161,163,163,161],14.877,[168,161,277,422,163,161,163,163,161],1.466,[171,161,277,424,163,161,163,163,161],0.574,[277,161,277,426,163,161,163,163,161],0.561,[282,161,277,428,163,161,163,163,161],0.577,[125,161,277,426,163,161,163,163,161],[291,161,277,431,163,161,163,163,161],12.913,[296,161,277,433,163,161,163,163,161],1.752,[161,161,282,435,163,161,163,163,161],4.255,[165,161,282,435,163,161,163,163,161],[168,161,282,438,163,161,163,163,161],5.735,[171,161,282,440,163,161,163,163,161],4.468,[277,161,282,442,163,161,163,163,161],4.152,[282,161,282,444,163,161,163,163,161],4.247,[125,161,282,446,163,161,163,163,161],4.644,[291,161,282,448,163,161,163,163,161],4.299,[296,161,282,450,163,161,163,163,161],5.588,[452],"failed",[315],[],[],[457],"KITTI sequences split into overlapping sessions; FAST-LIO2 initial odometry; FAST-LIO2 column is the single-run result; 'Fail' = merging failed; ablations without M-detector or loop filter, and variants replacing BTC by SOLiD or the RANSAC filter by PCM; the 'Origin' per-session column is not extracted",{"slug":459,"group":460,"sourceId":5,"sourceLabel":6,"table":132,"selfRows":125,"metrics":461,"seqs":463,"entrants":479,"cells":486,"outcomes":499,"locators":500,"hardware":501,"wordings":502,"notes":503},"lamm2025-table-v","lamm2025:Table V",[462],{"label":224,"unit":137,"statistic":225,"alignment":113},[464,468,470,472,475,477],{"dataset":465,"sequence":466,"environment":467},"HeLiPR","Town Ouster 1-3 merged","Town (residential area)",{"dataset":465,"sequence":469,"environment":467},"Town Avia 1-3 merged",{"dataset":465,"sequence":471,"environment":467},"Town all six Ouster and Avia sequences",{"dataset":465,"sequence":473,"environment":474},"Roundabout Ouster 1-3 merged","Roundabout (residential area)",{"dataset":465,"sequence":476,"environment":474},"Roundabout Avia 1-3 merged",{"dataset":465,"sequence":478,"environment":474},"Roundabout all six Ouster and Avia sequences",[480,482,484],{"name":481,"methodId":5,"linkable":156,"proposed":156,"self":156},"LAMM merged (Ouster sequences)",{"name":483,"methodId":5,"linkable":156,"proposed":156,"self":156},"LAMM merged (Avia sequences)",{"name":485,"methodId":5,"linkable":156,"proposed":156,"self":156},"LAMM merged multi-LiDAR",[487,489,491,493,495,497],[161,161,161,488,163,161,163,163,161],2.818,[165,161,165,490,163,161,163,163,161],4.273,[168,161,168,492,163,161,163,163,161],4.008,[161,161,171,494,163,161,163,163,161],2.125,[165,161,277,496,163,161,163,163,161],2.247,[168,161,282,498,163,161,163,163,161],2.101,[],[132],[],[],[504],"HeLiPR Town and Roundabout: FAST-LIO2 ATE of each input sequence, ATE after merging the three sequences of the same LiDAR type, and after merging all six Ouster and Avia sequences (single value per environment)",[506,510],{"group":507,"slug":508,"sourceLabel":6,"table":220,"selfRows":277,"datasets":509},"lamm2025:Table IV","lamm2025-table-iv",[465],{"group":511,"slug":512,"sourceLabel":131,"table":513,"selfRows":168,"datasets":514},"lemon2026:Table VI","lemon2026-table-vi","Table VI",[228],1790510654961]