[{"data":1,"prerenderedAt":538},["ShallowReactive",2],{"method-lim2024quatropp":3},{"method":4,"reference":63,"equipment":85,"figures":124,"results":125},{"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":28,"sensors":36,"platform":39,"estimator":43,"association":44,"timeModel":45,"deskew":46,"loopClosure":47,"globalOptimization":48,"mapRepresentation":49,"prior":50,"outputGeometry":51,"compute":52,"codeUrl":53,"codeLicense":54,"relatedVersions":55},"lim2024quatropp","Lim et al., 2024","Quatro++","Quatro++: Robust global registration exploiting ground segmentation for loop closing in LiDAR SLAM",2024,"recent","C02","registration_component","Quatro++ 針對 LiDAR SLAM 迴圈閉合中的全域配準，處理機械旋轉式 LiDAR 點雲稀疏、以及離群剔除後剩下不足三個內點造成退化兩個問題。方法先以地面分割移除幾何資訊少的地面點，再做特徵匹配與最大團內點選擇，並假設地面載具以偏航旋轉為主，以 GNC 估計準 SO(3) 旋轉與分量式平移；滾轉與俯仰可由 INS 補償。作者在 KITTI、NAVER LABS、MulRan 與手持式 HiltiOxford 資料上評估，並接入 SLAM 迴圈模組。","Quatro++ uses ground segmentation and a yaw-dominant quasi-SE(3) robust solver to improve global registration success for LiDAR loop closing under sparsity and degeneracy.","full_text_reviewed","peer_reviewed_published","background","HiltiOxford 手持資料只用於定性可行性展示（Fig. 18），且需先以 INS 補償滾轉與俯仰；論文沒有描述該資料的場域是否為工地。其餘評估為 KITTI 與 MulRan 車載資料以及 NAVER LABS 室內資料。地面接觸與偏航主導的假設能否用於多樓層或斜坡工地屬推論。",[20],"public_benchmark",[22,23,24,25,26,27],"higher success rate than state-of-the-art global registration under sparsity and degeneracy (abstract)","ground segmentation significantly increases success for ground vehicles (abstract)","improved loop-constraint quality and mapping precision (abstract, Sec. 8)","QSC-LeGO-LOAM gives the lowest APE mean on DCC01, KAIST02 and Riverside01 (5.65, 3.88, 20.19) (Table 7)","Quatro++-c2f reaches trel 0.50% and rrel 0.20 deg\u002F100m on KITTI Seq. 00 at Delta = 5 (Table 6)","handheld registration succeeds on HiltiOxford after INS roll-pitch compensation (Sec. 7.5, Fig. 18)",[29,30,31,32,33,34,35],"sacrifices relative roll\u002Fpitch estimation, requiring INS to recover (Sec. 5.5)","MSE-based false-loop rejection can cause false negatives and false positives (Sec. 5.3)","relies on ground contact and yaw-dominant motion assumptions (Sec. 4.1)","some failure cases with low MSE occur in corridor-like scenes, making false positive loops hard to reject (Sec. 5.3)","front-end feature extraction and matching are not improved; left to future work (Sec. 8)","at small frame intervals Quatro without ground segmentation had lower errors than Quatro++ (Sec. 7.3, Table 6)","ground segmentation occasionally lowers the success rate slightly (Sec. 7.2, Table 4)",[37,38],"3D LiDAR (Velodyne HDL-64E, VLP-16, Ouster OS1-64, HESAI XT32 across datasets)","INS optional for roll\u002Fpitch (Sec. 5.5)",[40,41,42],"vehicle (KITTI, MulRan)","handheld (HiltiOxford, qualitative feasibility only)","indoor NAVER LABS localization data (platform not stated)","decoupled estimation on translation-invariant measurements: quasi-SO(3) (yaw-only) rotation by GNC truncated least squares with alternating weight updates (noise bound 0.3, at most 50 iterations, kappa = 1.4), then component-wise translation estimation (COTE); the c2f variant adds G-ICP fine alignment","Patchwork ground segmentation removes ground points; voxel sampling; FPFH descriptors with sensor-specific radii (nu \u003C r_normal \u003C r_FPFH, Table 1); reciprocal-test matching; MCIS-heuristic pruning that keeps the maximal clique found within a time threshold","not_applicable","not_reported","coarse alignment in the loop-closing module of LeGO-LOAM; QSC-LeGO-LOAM combines ScanContext loop detection, Quatro++ and local registration, with MSE-based false-loop rejection","pose graph optimization over odometry and loop constraints (Eq. 14) in LeGO-LOAM","LiDAR scans","ground-contact assumption; optional INS roll\u002Fpitch","relative pose (quasi-SE(3)) for loop constraints","whole Quatro++ (preprocessing, correspondence estimation and Quatro) under 1 s per registration; Quatro optimization averages 5.0 ms (KITTI) and 6.4 ms (NAVER LABS) on an Intel Core i9-9900KF; preprocessing and matching times on Intel Core i7-7700K and i9-13900 are shown only as plots","https:\u002F\u002Fgithub.com\u002Furl-kaist\u002FQuatro","CC BY-NC-SA 4.0 (declared in README 'License' section; no LICENSE file; non-commercial terms)",[56,60],{"relation":57,"title":58,"doi_or_url":59},"conference_version","A Single Correspondence Is Enough: Robust Global Registration to Avoid Degeneracy in Urban Environments (ICRA 2022, pp. 8010-8017; arXiv:2203.06612)","10.1109\u002FICRA46639.2022.9812018",{"relation":61,"title":62,"doi_or_url":53},"code_release","Quatro (re-implementation; also integrated in TEASER++ per README)",{"id":5,"kind":64,"shortName":7,"title":8,"authors":65,"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":45,"codeUrl":53,"cluster":11,"topics":80,"mdpi":81,"verification":82,"label":6,"fulltextRoute":83,"versionRead":84,"addedByCensus":81},"method",[66,67,68,69,70],"Hyungtae Lim","Beomsoo Kim","Daebeom Kim","Eungchang Mason Lee","Hyun Myung","The International Journal of Robotics Research","journal","SAGE","43(5):685-715","10.1177\u002F02783649231207654","2311.00928","https:\u002F\u002Farxiv.org\u002Fabs\u002F2311.00928","2023-11-02","metadata_verified",[11],false,"corrected","arXiv","arXiv 2311.00928v2 (22 Jan 2024), SAGE preprint layout; IJRR version of record not compared",[86,92,95,98,104,111,118,120],{"category":87,"model":88,"canonical":88,"role":89,"dataset":90,"specs":46,"locator":91},"lidar","Velodyne HDL-64E","dataset sensor","KITTI","Sec. 6.1",{"category":87,"model":93,"canonical":93,"role":89,"dataset":94,"specs":46,"locator":91},"Velodyne VLP-16","NAVER LABS localization dataset",{"category":87,"model":96,"canonical":96,"role":89,"dataset":97,"specs":46,"locator":91},"Ouster OS1-64","MulRan",{"category":87,"model":99,"canonical":100,"role":89,"dataset":101,"specs":102,"locator":103},"HESAI XT32","Hesai XT-32","Hilti-Oxford (HiltiOxford)","hand-held sensor configuration","Sec. 6.1; Sec. 7.5",{"category":105,"model":106,"canonical":106,"role":107,"dataset":108,"specs":109,"locator":110},"imu","INS (model not stated)","method input","KITTI Seq. 06 (Table 5); HiltiOxford (Fig. 18)","raw roll and pitch used to compensate the source cloud before quasi-SO(3) estimation","Sec. 5.5; Sec. 7.5; Table 5",{"category":112,"model":113,"canonical":113,"role":114,"dataset":115,"specs":116,"locator":117},"compute","Intel Core i7-7700K","compute for runtime",null,"preprocessing and correspondence timing","Fig. 16",{"category":112,"model":119,"canonical":119,"role":114,"dataset":115,"specs":116,"locator":117},"Intel Core i9-13900",{"category":112,"model":121,"canonical":121,"role":114,"dataset":115,"specs":122,"locator":123},"Intel Core i9-9900KF","optimization timing (Quatro 5.0 ms KITTI, 6.4 ms NAVER LABS)","Fig. 17",[],{"totalRows":126,"groupCount":127,"groups":128,"others":537},25,3,[129,378,513],{"slug":130,"group":131,"sourceId":5,"sourceLabel":6,"table":132,"selfRows":133,"metrics":134,"seqs":141,"entrants":149,"cells":191,"outcomes":371,"locators":373,"hardware":374,"wordings":375,"notes":376},"lim2024quatropp-table-6","lim2024quatropp:Table 6","Table 6",12,[135,138],{"label":136,"unit":137,"statistic":46,"alignment":46},"trel","%",{"label":139,"unit":140,"statistic":46,"alignment":46},"rrel","deg\u002F100m",[142,145,147],{"dataset":90,"sequence":143,"environment":144},"Seq. 00, Delta = 1","vehicle, urban driving (Velodyne HDL-64E)",{"dataset":90,"sequence":146,"environment":144},"Seq. 00, Delta = 3",{"dataset":90,"sequence":148,"environment":144},"Seq. 00, Delta = 5",[150,154,157,160,163,166,168,170,173,175,177,179,181,184,187,189],{"name":151,"methodId":152,"linkable":153,"proposed":81,"self":81},"ICP","besl1992icp",true,{"name":155,"methodId":156,"linkable":153,"proposed":81,"self":81},"G-ICP","segal2009gicp",{"name":158,"methodId":159,"linkable":153,"proposed":81,"self":81},"VGICP","koide2021vgicp",{"name":161,"methodId":162,"linkable":153,"proposed":81,"self":81},"FGR","zhou2016fgr",{"name":164,"methodId":165,"linkable":153,"proposed":81,"self":81},"TEASER++","yang2021teaser",{"name":167,"methodId":115,"linkable":81,"proposed":153,"self":81},"Quatro (Ours)",{"name":169,"methodId":5,"linkable":153,"proposed":153,"self":153},"Quatro++ (Ours)",{"name":171,"methodId":172,"linkable":153,"proposed":81,"self":81},"LO-Net","lonet2019",{"name":174,"methodId":172,"linkable":153,"proposed":81,"self":81},"LO-Net+M",{"name":176,"methodId":115,"linkable":81,"proposed":81,"self":81},"DMLO†",{"name":178,"methodId":115,"linkable":81,"proposed":81,"self":81},"DMLO+M†",{"name":180,"methodId":115,"linkable":81,"proposed":81,"self":81},"A-LOAM + StickyPillars†",{"name":182,"methodId":183,"linkable":153,"proposed":81,"self":81},"SuMa","suma2018",{"name":185,"methodId":186,"linkable":153,"proposed":81,"self":81},"A-LOAM","aloam_software",{"name":188,"methodId":115,"linkable":81,"proposed":153,"self":81},"Quatro-c2f (Ours)",{"name":190,"methodId":5,"linkable":153,"proposed":153,"self":153},"Quatro++-c2f (Ours)",[192,196,199,201,203,206,208,210,212,214,216,218,220,222,224,226,228,230,232,234,236,238,240,242,244,247,249,251,253,255,256,258,260,262,264,266,268,271,273,274,276,278,280,283,285,286,287,288,289,292,294,295,296,297,298,301,303,304,305,306,307,310,311,312,313,314,315,318,320,322,324,326,328,330,332,334,336,338,340,343,345,347,349,351,353,355,357,359,360,361,362,364,365,366,367,369],[193,193,193,194,195,193,195,195,193],0,6.88,-1,[193,197,193,198,195,193,195,195,193],1,2.99,[193,193,197,200,195,193,195,195,193],21.92,[193,197,197,202,195,193,195,195,193],8.7,[193,193,204,205,195,193,195,195,193],2,21.14,[193,197,204,207,195,193,195,195,193],8.51,[197,193,193,209,195,193,195,195,193],1.26,[197,197,193,211,195,193,195,195,193],0.45,[197,193,197,213,195,193,195,195,193],5.5,[197,197,197,215,195,193,195,195,193],1.45,[197,193,204,217,195,193,195,195,193],14.2,[197,197,204,219,195,193,195,195,193],3.32,[204,193,193,221,195,193,195,195,193],1.03,[204,197,193,223,195,193,195,195,193],0.3,[204,193,197,225,195,193,195,195,193],11.83,[204,197,197,227,195,193,195,195,193],1.65,[204,193,204,229,195,193,195,195,193],19.11,[204,197,204,231,195,193,195,195,193],6.32,[127,193,193,233,195,193,195,195,193],2.73,[127,197,193,235,195,193,195,195,193],0.69,[127,193,197,237,195,193,195,195,193],7.17,[127,197,197,239,195,193,195,195,193],1.58,[127,193,204,241,195,193,195,195,193],14.66,[127,197,204,243,195,193,195,195,193],4.12,[245,193,193,246,195,193,195,195,193],4,2.11,[245,197,193,248,195,193,195,195,193],0.91,[245,193,197,250,195,193,195,195,193],2.64,[245,197,197,252,195,193,195,195,193],1.11,[245,193,204,254,195,193,195,195,193],3.19,[245,197,204,248,195,193,195,195,193],[257,193,193,215,195,193,195,195,193],5,[257,197,193,259,195,193,195,195,193],0.41,[257,193,197,261,195,193,195,195,193],1.38,[257,197,197,263,195,193,195,195,193],0.24,[257,193,204,265,195,193,195,195,193],1.94,[257,197,204,267,195,193,195,195,193],0.46,[269,193,193,270,195,193,195,195,193],6,1.9,[269,197,193,272,195,193,195,195,193],0.53,[269,193,197,215,195,193,195,195,193],[269,197,197,275,195,193,195,195,193],0.32,[269,193,204,277,195,193,195,195,193],0.99,[269,197,204,279,195,193,195,195,193],0.28,[281,193,193,282,195,193,195,195,193],7,1.47,[281,197,193,284,195,193,195,195,193],0.72,[281,193,197,115,193,193,195,195,193],[281,197,197,115,193,193,195,195,193],[281,193,204,115,193,193,195,195,193],[281,197,204,115,193,193,195,195,193],[290,193,193,291,195,193,195,195,193],8,0.78,[290,197,193,293,195,193,195,195,193],0.42,[290,193,197,115,193,193,195,195,193],[290,197,197,115,193,193,195,195,193],[290,193,204,115,193,193,195,195,193],[290,197,204,115,193,193,195,195,193],[299,193,193,300,195,193,195,195,193],9,0.83,[299,197,193,302,195,193,195,195,193],0.44,[299,193,197,115,193,193,195,195,193],[299,197,197,115,193,193,195,195,193],[299,193,204,115,193,193,195,195,193],[299,197,204,115,193,193,195,195,193],[308,193,193,309,195,193,195,195,193],10,0.73,[308,197,193,302,195,193,195,195,193],[308,193,197,115,193,193,195,195,193],[308,197,197,115,193,193,195,195,193],[308,193,204,115,193,193,195,195,193],[308,197,204,115,193,193,195,195,193],[316,193,193,317,195,193,195,195,193],11,0.65,[316,197,193,319,195,193,195,195,193],0.26,[316,193,197,321,195,193,195,195,193],0.79,[316,197,197,323,195,193,195,195,193],0.31,[316,193,204,325,195,193,195,195,193],1.29,[316,197,204,327,195,193,195,195,193],0.48,[133,193,193,329,195,193,195,195,193],0.68,[133,197,193,331,195,193,195,195,193],0.23,[133,193,197,333,195,193,195,195,193],1.69,[133,197,197,335,195,193,195,195,193],0.61,[133,193,204,337,195,193,195,195,193],2.36,[133,197,204,339,195,193,195,195,193],0.51,[341,193,193,342,195,193,195,195,193],13,0.7,[341,197,193,344,195,193,195,195,193],0.27,[341,193,197,346,195,193,195,195,193],0.97,[341,197,197,348,195,193,195,195,193],0.38,[341,193,204,350,195,193,195,195,193],31.16,[341,197,204,352,195,193,195,195,193],12.1,[354,193,193,317,195,193,195,195,193],14,[354,197,193,356,195,193,195,195,193],0.21,[354,193,197,358,195,193,195,195,193],0.67,[354,197,197,356,195,193,195,195,193],[354,193,204,358,195,193,195,195,193],[354,197,204,356,195,193,195,195,193],[363,193,193,329,195,193,195,195,193],15,[363,197,193,331,195,193,195,195,193],[363,193,197,335,195,193,195,195,193],[363,197,197,356,195,193,195,195,193],[363,193,204,368,195,193,195,195,193],0.5,[363,197,204,370,195,193,195,195,193],0.2,[372],"not_reported (N\u002FA in table; not available from the original paper)",[132],[],[],[377],"KITTI Seq. 00 odometry test with frame interval Delta (source i+Delta, target i); trel [%] and rrel [deg\u002F100m] by RPG evaluation tools; c2f = global registration then local registration (G-ICP); deep-learning rows copied by the authors from the original papers; † = Seq. 00 used for training",{"slug":379,"group":380,"sourceId":5,"sourceLabel":6,"table":381,"selfRows":133,"metrics":382,"seqs":396,"entrants":404,"cells":414,"outcomes":507,"locators":508,"hardware":509,"wordings":510,"notes":511},"lim2024quatropp-table-7","lim2024quatropp:Table 7","Table 7",[383,387,390,393],{"label":384,"unit":385,"statistic":386,"alignment":46},"Absolute pose error, Mean","not_stated","mean",{"label":388,"unit":385,"statistic":389,"alignment":46},"Absolute pose error, Median","median",{"label":391,"unit":385,"statistic":392,"alignment":46},"Absolute pose error, RMSE","RMSE",{"label":394,"unit":385,"statistic":395,"alignment":46},"Absolute pose error, Stdev.","std",[397,400,402],{"dataset":97,"sequence":398,"environment":399},"DCC01","vehicle, large-scale outdoor sequences DCC01, KAIST02 and Riverside01 (scene types not described beyond 'large-scale' and 'riverside scenes')",{"dataset":97,"sequence":401,"environment":399},"KAIST02",{"dataset":97,"sequence":403,"environment":399},"Riverside01",[405,408,410,412],{"name":406,"methodId":407,"linkable":153,"proposed":81,"self":81},"LeGO-LOAM","legoloam2018",{"name":409,"methodId":115,"linkable":81,"proposed":81,"self":81},"SC-LeGO-LOAM",{"name":411,"methodId":115,"linkable":81,"proposed":81,"self":81},"TSC-LeGO-LOAM",{"name":413,"methodId":5,"linkable":153,"proposed":153,"self":153},"QSC-LeGO-LOAM (Ours)",[415,417,419,421,423,425,427,429,431,433,435,437,439,441,443,445,447,449,451,453,455,457,459,460,462,464,466,468,470,472,474,476,478,480,482,484,486,488,490,492,493,495,496,498,500,502,504,506],[193,193,193,416,195,193,195,195,193],28.89,[193,197,193,418,195,193,195,195,193],29.21,[193,204,193,420,195,193,195,195,193],32.37,[193,127,193,422,195,193,195,195,193],14.6,[197,193,193,424,195,193,195,195,193],6.13,[197,197,193,426,195,193,195,195,193],5.59,[197,204,193,428,195,193,195,195,193],6.46,[197,127,193,430,195,193,195,195,193],1.89,[204,193,193,432,195,193,195,195,193],6.1,[204,197,193,434,195,193,195,195,193],5.49,[204,204,193,436,195,193,195,195,193],5.56,[204,127,193,438,195,193,195,195,193],2.58,[127,193,193,440,195,193,195,195,193],5.65,[127,197,193,442,195,193,195,195,193],5.33,[127,204,193,444,195,193,195,195,193],5.97,[127,127,193,446,195,193,195,195,193],1.92,[193,193,197,448,195,193,195,195,193],24.21,[193,197,197,450,195,193,195,195,193],18.48,[193,204,197,452,195,193,195,195,193],30.25,[193,127,197,454,195,193,195,195,193],18.14,[197,193,197,456,195,193,195,195,193],4.28,[197,197,197,458,195,193,195,195,193],3.72,[197,204,197,442,195,193,195,195,193],[197,127,197,461,195,193,195,195,193],3.17,[204,193,197,463,195,193,195,195,193],4.27,[204,197,197,465,195,193,195,195,193],3.85,[204,204,197,467,195,193,195,195,193],5.12,[204,127,197,469,195,193,195,195,193],2.82,[127,193,197,471,195,193,195,195,193],3.88,[127,197,197,473,195,193,195,195,193],2.71,[127,204,197,475,195,193,195,195,193],4.86,[127,127,197,477,195,193,195,195,193],2.93,[193,193,204,479,195,193,195,195,193],96.6,[193,197,204,481,195,193,195,195,193],62.57,[193,204,204,483,195,193,195,195,193],124.94,[193,127,204,485,195,193,195,195,193],79.27,[197,193,204,487,195,193,195,195,193],26.87,[197,197,204,489,195,193,195,195,193],28.99,[197,204,204,491,195,193,195,195,193],30.59,[197,127,204,487,195,193,195,195,193],[204,193,204,494,195,193,195,195,193],21.23,[204,197,204,494,195,193,195,195,193],[204,204,204,497,195,193,195,195,193],32.89,[204,127,204,499,195,193,195,195,193],19.5,[127,193,204,501,195,193,195,195,193],20.19,[127,197,204,503,195,193,195,195,193],19.57,[127,204,204,505,195,193,195,195,193],23.01,[127,127,204,503,195,193,195,195,193],[],[381],[],[],[512],"absolute pose errors of full SLAM results on MulRan (Ouster OS1-64, vehicle); LeGO-LOAM variants: SC = ScanContext loop detection, TSC = TEASER++ + ScanContext, QSC = Quatro++ + ScanContext; unit not stated (presumably m); values as printed (for TSC on DCC01 RMSE \u003C mean, and SC on Riverside01 std = mean)",{"slug":514,"group":515,"sourceId":5,"sourceLabel":6,"table":516,"selfRows":197,"metrics":517,"seqs":521,"entrants":524,"cells":527,"outcomes":529,"locators":531,"hardware":533,"wordings":534,"notes":535},"lim2024quatropp-text-sec-7-4","lim2024quatropp:Text Sec. 7.4","Text Sec. 7.4",[518],{"label":519,"unit":520,"statistic":46,"alignment":45},"total time of Quatro++ takes less than one second","s",[522],{"dataset":523,"sequence":46,"environment":523},"not stated",[525],{"name":526,"methodId":5,"linkable":153,"proposed":153,"self":153},"Quatro++ (whole pipeline)",[528],[193,193,193,197,193,193,195,195,193],[530],"upper bound (less than 1 s per registration)",[532],"Sec. 7.4",[],[],[536],"total time of Quatro++ (preprocessing, correspondence estimation and Quatro) stated as an upper bound; hardware not named for the total",[],1790510659517]