[{"data":1,"prerenderedAt":328},["ShallowReactive",2],{"method-std2023":3},{"method":4,"reference":56,"equipment":78,"figures":104,"results":105},{"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":28,"sensors":36,"platform":38,"estimator":41,"association":42,"timeModel":43,"deskew":44,"loopClosure":45,"globalOptimization":46,"mapRepresentation":47,"prior":48,"outputGeometry":43,"compute":49,"codeUrl":50,"codeLicense":51,"relatedVersions":52},"std2023","Yuan et al., 2023b","STD","STD: Stable Triangle Descriptor for 3D place recognition",2023,"recent","C06","place_recognition_component","STD 在由數次掃描累積而成的關鍵影格上，先以體素共變異數矩陣的特徵值判斷平面並以區域成長擴展，再把平面邊界體素中的點投影到所屬平面形成影像，取 5×5 鄰域極大值作為關鍵點；每個關鍵點以 kd-tree 取 20 個近鄰組成三角形，三邊長與三個法向量內積共六個屬性對剛體變換不變，作為雜湊鍵投票檢索前 10 個候選關鍵影格；再以三角形頂點對應經 SVD 與 RANSAC 求相對位姿，並以平面重合比例做幾何驗證，可選擇以 STD-ICP 精化位姿。方法支援非重複掃描的固態光達。","STD extracts plane-boundary keypoints from accumulated keyframes, encodes them as rigid-invariant triangles stored in a hash table, and verifies loops geometrically, supporting non-repetitive solid-state LiDARs.","full_text_reviewed","peer_reviewed_published","main_body","無工地測試；作者以 Livox Avia 蒐集多樓層建築室內資料，指出各樓層走廊相似導致精確率與召回率較低，但仍能提供一定數量的有效迴圈，作者認為可用於多樓層停車場、博物館等室內建圖（Sec. IV-B1）。營建中重複樓層與標準化空間為直接相關風險（推論）；LTA-OM 以 STD 為迴圈偵測並在多層相似結構建築驗證（ltaom2024 abstract）。",[20,21],"public_benchmark","completed_building",[23,24,25,26,27],"Works on spinning and non-repetitive solid-state LiDARs with the same parameters except indoor voxel size (Sec. IV-A, IV-B)","Descriptor matching gives point correspondences for geometric verification and relative pose (abstract; Sec. III)","Gives a full 6-DoF relative pose; on KITTI00 loop nodes with perturbed initial values, STD-ICP reached accuracy similar to GICP with lower variance at less than 1% of GICP time (Sec. IV-A4, Fig. 10)","Hash-table database keeps query time from growing linearly with database size, unlike Scan Context and M2DP (Sec. IV-A2, Fig. 8)","Detected loops in narrow scenes with little vertical variation where Scan Context is reported to struggle (Sec. IV-A1, Fig. 7a)",[29,30,31,32,33,34,35],"Indoor multi-floor building dataset showed relatively low precision and recall because corridors on each floor are very similar (Sec. IV-B1)","Sec. IV-A states that Scan Context and M2DP results on the public datasets were taken directly from the Scan Context paper, so conditions are not identical, while Sec. IV-A2 says the modified Scan Context MATLAB code (8 augmentations) was used to obtain the Sec. IV-A1 results; the source of the baseline curves is therefore internally inconsistent (Sec. IV-A; Sec. IV-A2)","Ground truth for the Livox datasets was built from LiDAR-inertial odometry plus loop closure and pose graph, not an independent reference (Sec. IV-B)","Authors state STD performs poorly only when structures or planes are particularly sparse because extracted keypoints become scarce (Sec. IV-A1, Fig. 7b, NCLT)","Scan Context was not compared on the Livox datasets because the authors consider it incompatible with Livox solid-state LiDARs; indoor true-positive threshold was 4 m versus 20 m outdoors (Sec. IV-B)","Runtime baselines ran from MATLAB code (M2DP default, Scan Context modified with 8 augmentations) while the STD implementation language is not stated, so timing parity is unclear (Sec. IV-A2; parity point is an inference)","Results are reported only as plots (precision-recall, time, error), with no tabulated values (Figs. 6, 8-10, 12)",[37],"3D LiDAR",[39,40],"vehicle","not_verified","not_applicable for odometry; the loop relative pose is solved in closed form by SVD from the three matched triangle vertices inside RANSAC (maximizing correctly matched descriptors), optionally refined by STD-ICP, a Ceres optimization of plane normal difference and point-to-plane distance between coinciding planes","triangle descriptors from keypoints on plane boundaries; hash table on rotation\u002Ftranslation-invariant side lengths and normal dot products; top-10 candidates; RANSAC and plane-based geometric verification","not_applicable","not_reported (motion compensation is not discussed); the component receives scans already registered by an external LiDAR odometry and accumulates 10 scans per keyframe for spinning LiDARs and 20 for Livox LiDARs","hash-table voting selects the top-10 candidate keyframes; a candidate is accepted when the plane coincidence percentage after the RANSAC transform exceeds sigma_pc (0.5 to 0.6 suggested from KITTI08); returns a 6-DoF relative pose, with loop correction left to the host SLAM back-end","none (component)","per-keyframe triangle descriptor hash database plus extracted planes","none; relies on an external LiDAR odometry to register scans into keyframes (text cites ref. [28], the adaptive voxel map odometry of Yuan et al. 2022; the Fig. 2 input block is labelled LiDAR odometry and mapping (LOAM))","All experiments run on one system with an Intel i7-11700k @ 3.6 GHz and 16 GB memory (no GPU is mentioned); per-frame time does not grow linearly with the number of stored frames because of the hash table, and is similar to M2DP while processing 10 times more points on KITTI00; STD and STD-ICP take less than 1% of GICP time for loop-node registration","https:\u002F\u002Fgithub.com\u002Fhku-mars\u002FSTD","GPL-2.0 per README, free only for personal and academic use; commercial use requires negotiation",[53],{"relation":54,"title":55,"doi_or_url":50},"code_release","hku-mars\u002FSTD",{"id":5,"kind":57,"shortName":7,"title":8,"authors":58,"year":9,"venue":64,"venueType":65,"publisher":66,"volumeIssuePages":67,"doi":68,"arxivId":69,"url":70,"firstPublicDate":71,"publicationStatus":16,"metadataStatus":72,"fulltextStatus":15,"era":10,"classicReason":43,"codeUrl":50,"cluster":11,"topics":73,"mdpi":74,"verification":75,"label":6,"fulltextRoute":76,"versionRead":77,"addedByCensus":74},"component",[59,60,61,62,63],"Chongjian Yuan","Jiarong Lin","Zuhao Zou","Xiaoping Hong","Fu Zhang","2023 IEEE International Conference on Robotics and Automation (ICRA)","conference","IEEE","pp. 1897-1903","10.1109\u002Ficra48891.2023.10160413","2209.12435","https:\u002F\u002Farxiv.org\u002Fabs\u002F2209.12435","2022-09-26","metadata_verified",[11],false,"corrected","arXiv","arXiv v2 (2023-02-22, 7 pages incl. references); ICRA 2023 IEEE version of record (pp. 1897-1903) not read",[79,86,92,97],{"category":80,"model":81,"canonical":81,"role":82,"dataset":83,"specs":84,"locator":85},"lidar","Livox Avia","method input","self-collected Avia Park1, Avia Park2 and Avia Indoor (multi-floor building)","small-FOV, non-repetitive scanning solid-state LiDAR; 20 frames accumulated per keyframe","Abstract; Fig. 1; Sec. IV-B; Fig. 12",{"category":80,"model":87,"canonical":87,"role":88,"dataset":89,"specs":90,"locator":91},"Livox Horizon","dataset sensor","KA Urban East (open-sourced with LiLi-OM)","solid-state LiDAR","Sec. IV-B",{"category":80,"model":93,"canonical":93,"role":88,"dataset":94,"specs":95,"locator":96},"mechanical spinning LiDARs (models not reported)","KITTI odometry, NCLT, Complex Urban","different numbers of scanning lines; 10 frames accumulated per keyframe","Sec. IV-A",{"category":98,"model":99,"canonical":99,"role":100,"dataset":101,"specs":102,"locator":103},"compute","Intel i7-11700k","compute for runtime",null,"@ 3.6 GHz with 16 GB memory; same system for all experiments","Sec. IV",[],{"totalRows":106,"groupCount":107,"groups":108,"others":327},9,1,[109],{"slug":110,"group":111,"sourceId":112,"sourceLabel":113,"table":114,"selfRows":106,"metrics":115,"seqs":128,"entrants":133,"cells":183,"outcomes":321,"locators":322,"hardware":323,"wordings":324,"notes":325},"lim2025kissmatcher-table-i","lim2025kissmatcher:Table I","lim2025kissmatcher","Lim et al., 2025","Table I",[116,121,124],{"label":117,"unit":118,"statistic":119,"alignment":120},"RTE [cm]","cm","mean","none",{"label":122,"unit":123,"statistic":119,"alignment":120},"RRE [deg]","deg",{"label":125,"unit":126,"statistic":127,"alignment":120},"Success rate [%]","%","not_reported",[129],{"dataset":130,"sequence":131,"environment":132},"KITTI","10 m benchmark","vehicle, urban driving",[134,136,140,142,144,146,148,150,153,155,157,159,161,163,165,168,171,173,175,177,179,181],{"name":135,"methodId":101,"linkable":74,"proposed":74,"self":74},"3DFeat-Net",{"name":137,"methodId":138,"linkable":139,"proposed":74,"self":74},"FCGF","choy2019fcgf",true,{"name":141,"methodId":101,"linkable":74,"proposed":74,"self":74},"DIP",{"name":143,"methodId":101,"linkable":74,"proposed":74,"self":74},"Predator",{"name":145,"methodId":101,"linkable":74,"proposed":74,"self":74},"SpinNet",{"name":147,"methodId":101,"linkable":74,"proposed":74,"self":74},"D3Feat",{"name":149,"methodId":101,"linkable":74,"proposed":74,"self":74},"GeDi",{"name":151,"methodId":152,"linkable":139,"proposed":74,"self":74},"G-ICP","segal2009gicp",{"name":154,"methodId":5,"linkable":139,"proposed":74,"self":139},"STD, W = 1",{"name":156,"methodId":5,"linkable":139,"proposed":74,"self":139},"STD, W = 3",{"name":158,"methodId":5,"linkable":139,"proposed":74,"self":139},"STD, W = 5",{"name":160,"methodId":101,"linkable":74,"proposed":74,"self":74},"MapClosures, W = 1",{"name":162,"methodId":101,"linkable":74,"proposed":74,"self":74},"MapClosures, W = 3",{"name":164,"methodId":101,"linkable":74,"proposed":74,"self":74},"MapClosures, W = 5",{"name":166,"methodId":167,"linkable":139,"proposed":74,"self":74},"FPFH + FGR","zhou2016fgr",{"name":169,"methodId":170,"linkable":139,"proposed":74,"self":74},"FPFH + TEASER++","yang2021teaser",{"name":172,"methodId":101,"linkable":74,"proposed":74,"self":74},"FPFH + Quatro",{"name":174,"methodId":112,"linkable":139,"proposed":139,"self":74},"Proposed",{"name":176,"methodId":101,"linkable":74,"proposed":74,"self":74},"FPFH + FGR + G-ICP",{"name":178,"methodId":101,"linkable":74,"proposed":74,"self":74},"FPFH + TEASER + G-ICP",{"name":180,"methodId":101,"linkable":74,"proposed":74,"self":74},"FPFH + Quatro + G-ICP",{"name":182,"methodId":112,"linkable":139,"proposed":139,"self":74},"Proposed + G-ICP",[184,188,190,193,195,197,199,201,203,205,208,210,211,214,216,218,221,222,223,226,228,229,232,234,236,239,241,243,245,246,248,251,253,255,257,259,261,264,266,268,271,272,274,277,278,280,283,285,287,290,292,293,296,297,299,302,304,306,309,311,312,314,316,317,319,320],[185,185,185,186,187,185,187,187,185],0,25.9,-1,[185,107,185,189,187,185,187,187,185],0.57,[185,191,185,192,187,185,187,187,185],2,95.97,[107,185,185,194,187,185,187,187,185],6.47,[107,107,185,196,187,185,187,187,185],0.23,[107,191,185,198,187,185,187,187,185],99.82,[191,185,185,200,187,185,187,187,185],8.69,[191,107,185,202,187,185,187,187,185],0.44,[191,191,185,204,187,185,187,187,185],97.3,[206,185,185,207,187,185,187,187,185],3,5.6,[206,107,185,209,187,185,187,187,185],0.24,[206,191,185,198,187,185,187,187,185],[212,185,185,213,187,185,187,187,185],4,9.88,[212,107,185,215,187,185,187,187,185],0.47,[212,191,185,217,187,185,187,187,185],99.1,[219,185,185,220,187,185,187,187,185],5,11,[219,107,185,209,187,185,187,187,185],[219,191,185,198,187,185,187,187,185],[224,185,185,225,187,185,187,187,185],6,7.55,[224,107,185,227,187,185,187,187,185],0.33,[224,191,185,198,187,185,187,187,185],[230,185,185,231,187,185,187,187,185],7,8.56,[230,107,185,233,187,185,187,187,185],0.22,[230,191,185,235,187,185,187,187,185],37.95,[237,185,185,238,187,185,187,187,185],8,26.09,[237,107,185,240,187,185,187,187,185],0.69,[237,191,185,242,187,185,187,187,185],19.6,[106,185,185,244,187,185,187,187,185],20.94,[106,107,185,189,187,185,187,187,185],[106,191,185,247,187,185,187,187,185],30.58,[249,185,185,250,187,185,187,187,185],10,23.97,[249,107,185,252,187,185,187,187,185],0.66,[249,191,185,254,187,185,187,187,185],33.09,[220,185,185,256,187,185,187,187,185],38.5,[220,107,185,258,187,185,187,187,185],1.09,[220,191,185,260,187,185,187,187,185],69.38,[262,185,185,263,187,185,187,187,185],12,31.5,[262,107,185,265,187,185,187,187,185],0.95,[262,191,185,267,187,185,187,187,185],82.08,[269,185,185,270,187,185,187,187,185],13,32.27,[269,107,185,107,187,185,187,187,185],[269,191,185,273,187,185,187,187,185],86.64,[275,185,185,276,187,185,187,187,185],14,6.94,[275,107,185,227,187,185,187,187,185],[275,191,185,279,187,185,187,187,185],98.92,[281,185,185,282,187,185,187,187,185],15,9.36,[281,107,185,284,187,185,187,187,185],0.59,[281,191,185,286,187,185,187,187,185],99.64,[288,185,185,289,187,185,187,187,185],16,13.15,[288,107,185,291,187,185,187,187,185],0.94,[288,191,185,286,187,185,187,187,185],[294,185,185,295,187,185,187,187,185],17,18.1,[294,107,185,291,187,185,187,187,185],[294,191,185,298,187,185,187,187,185],100,[300,185,185,301,187,185,187,187,185],18,1.22,[300,107,185,303,187,185,187,187,185],0.04,[300,191,185,305,187,185,187,187,185],99.28,[307,185,185,308,187,185,187,187,185],19,1.1,[307,107,185,310,187,185,187,187,185],0.02,[307,191,185,286,187,185,187,187,185],[313,185,185,308,187,185,187,187,185],20,[313,107,185,315,187,185,187,187,185],0.03,[313,191,185,286,187,185,187,187,185],[318,185,185,308,187,185,187,187,185],21,[318,107,185,310,187,185,187,187,185],[318,191,185,298,187,185,187,187,185],[],[114],[],[],[326],"KITTI 10 m benchmark [23]: scan-to-scan global registration; success if translation \u003C 2 m and rotation \u003C 5 deg; RTE and RRE averaged over successful registrations only (Sec. IV-A); W = submap window size; learning-based results as listed by the authors",[],1790510654827]