[{"data":1,"prerenderedAt":397},["ShallowReactive",2],{"method-lim2025kissmatcher":3},{"method":4,"reference":57,"equipment":81,"figures":99,"results":100},{"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":33,"platform":36,"estimator":39,"association":40,"timeModel":41,"deskew":41,"loopClosure":42,"globalOptimization":43,"mapRepresentation":44,"prior":45,"outputGeometry":46,"compute":47,"codeUrl":48,"codeLicense":49,"relatedVersions":50},"lim2025kissmatcher","Lim et al., 2025","KISS-Matcher","KISS-Matcher: Fast and Robust Point Cloud Registration Revisited",2025,"recent","C02","registration_component","KISS-Matcher 從整體流程角度重新設計全域點雲配準，組合幾何抑制（如地面分割）、改良自 FPFH 的 Faster-PFH 特徵、以 k-core 為基礎的圖論離群剔除（降低 TEASER++ 最大團搜尋的時間複雜度）與 GNC 求解器，並釋出開源 C++ 函式庫。作者在 KITTI、MulRan 迴圈閉合基準以及以 FAST-LIO2 產生的多機器人地圖層級點雲上測試，報告精度與先進方法相當但速度大幅提升，可由掃描擴展到地圖層級。","KISS-Matcher combines Faster-PFH, k-core outlier pruning and a GNC solver into an open-source global registration pipeline scaling from scans to maps.","full_text_reviewed","peer_reviewed_published","main_body","not_reported",[20],"public_benchmark",[22,23,24,25,26],"on par with state of the art on KITTI while much faster (Sec. V)","generalizes across datasets and scales without training (Sec. I, IV)","final inlier count allows rejecting failure cases (Fig. 5 caption)","100% success on the KITTI 10 m benchmark, alone and with G-ICP (Table I)","only compared method that succeeded in map-level registration of Kimera-Multi FAST-LIO2 maps (Sec. IV-D, Fig. 6)",[28,29,30,31,32],"evaluation centred on driving and multi-robot outdoor datasets; indoor or construction data not reported in sections read (reviewer observation)","application to mapping\u002Flocalization left for future work (Sec. V)","standalone RTE 18.10 cm and RRE 0.94 deg are higher than most compared methods; G-ICP fine alignment lowers RTE to 1.10 cm (Table I)","runtime still grows linearly with the number of correspondences, hence the N_tau cap (Sec. III-D)","(reviewer observation) map-level evaluation is qualitative only (Fig. 6)",[34,35],"64-channel 3D LiDAR (KITTI and MulRan, different ray patterns; models not named in the paper)","map clouds produced by FAST-LIO2-based SLAM on the Kimera-Multi dataset (sensors not stated in the paper)",[37,38],"vehicle (KITTI, MulRan)","multi-robot maps from the Kimera-Multi dataset (robot type not stated in the paper)","maximum k-core pruning of a pairwise-invariant compatibility graph (O(|V|+|E|), CSR storage, beta = 1.5v) followed by a GNC non-minimal solver; the final inlier count is used to reject failed registrations; all parameters scale with voxel size v (r_normal = 3.5v, r_FPFH = 5.0v)","geometric suppression (Patchwork ground segmentation in the experiments); Faster-PFH with one radius search per point, a linearity filter (tau_lin = 0.99) and minimum neighbour count (tau_num = 3); mutual (reciprocity) matching; top N_tau = 3,000 correspondences by descriptor distance ratio","not_applicable","evaluated on the loop-closing benchmark of Lim et al. (Sec. IV-A)","none","scan-, submap- and map-level point clouds (voxelized) (Sec. IV-A)","none (global registration without initial guess)","rigid transformation from scan to map level","entire pipeline about 14 Hz on an Intel Core i9-13900 versus about 6 Hz for other outlier-robust pipelines and about 0.1 Hz (9.57 s per pair) for Predator (Table I caption); Faster-PFH about 4.5x (single-thread) and 2.4x (multi-thread) faster than FPFH (Sec. III-C); more than 20x faster than the TEASER++ pipeline above 200K voxelized points (Sec. IV-E, Fig. 1(b))","https:\u002F\u002Fgithub.com\u002FMIT-SPARK\u002FKISS-Matcher","MIT (LICENSE file checked)",[51,55],{"relation":52,"title":53,"doi_or_url":54},"preprint","arXiv:2409.15615 (v1 2024-09-23; v3 2025-07-16)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2409.15615",{"relation":56,"title":7,"doi_or_url":48},"code_release",{"id":5,"kind":58,"shortName":7,"title":8,"authors":59,"year":9,"venue":68,"venueType":69,"publisher":70,"volumeIssuePages":71,"doi":72,"arxivId":73,"url":54,"firstPublicDate":74,"publicationStatus":16,"metadataStatus":75,"fulltextStatus":15,"era":10,"classicReason":41,"codeUrl":48,"cluster":11,"topics":76,"mdpi":77,"verification":78,"label":6,"fulltextRoute":79,"versionRead":80,"addedByCensus":77},"method",[60,61,62,63,64,65,66,67],"Hyungtae Lim","Daebeom Kim","Gunhee Shin","Jingnan Shi","Ignacio Vizzo","Hyun Myung","Jaesik Park","Luca Carlone","2025 IEEE International Conference on Robotics and Automation (ICRA)","conference","IEEE","pp. 11104-11111","10.1109\u002Ficra55743.2025.11127458","2409.15615","2024-09-23","metadata_verified",[11],false,"corrected","arXiv","arXiv 2409.15615v3 (16 Jul 2025) including Appendix I; IEEE ICRA 2025 version of record not compared",[82,89,92],{"category":83,"model":84,"canonical":84,"role":85,"dataset":86,"specs":87,"locator":88},"lidar","64-channel LiDAR sensor (model not named in the paper)","dataset sensor","KITTI","laser ray pattern differs from MulRan's","Sec. IV-D; Fig. 5 caption",{"category":83,"model":84,"canonical":84,"role":85,"dataset":90,"specs":91,"locator":88},"MulRan","laser ray pattern differs from KITTI's",{"category":93,"model":94,"canonical":94,"role":95,"dataset":96,"specs":97,"locator":98},"compute","Intel Core i9-13900","compute for runtime",null,"runtime and parameter studies","Table I caption; Sec. III-C; Fig. 7 caption",[],{"totalRows":101,"groupCount":102,"groups":103,"others":396},10,4,[104,318,349,374],{"slug":105,"group":106,"sourceId":5,"sourceLabel":6,"table":107,"selfRows":108,"metrics":109,"seqs":120,"entrants":124,"cells":175,"outcomes":312,"locators":313,"hardware":314,"wordings":315,"notes":316},"lim2025kissmatcher-table-i","lim2025kissmatcher:Table I","Table I",6,[110,114,117],{"label":111,"unit":112,"statistic":113,"alignment":43},"RTE [cm]","cm","mean",{"label":115,"unit":116,"statistic":113,"alignment":43},"RRE [deg]","deg",{"label":118,"unit":119,"statistic":18,"alignment":43},"Success rate [%]","%",[121],{"dataset":86,"sequence":122,"environment":123},"10 m benchmark","vehicle, urban driving",[125,127,131,133,135,137,139,141,144,147,149,151,153,155,157,160,163,165,167,169,171,173],{"name":126,"methodId":96,"linkable":77,"proposed":77,"self":77},"3DFeat-Net",{"name":128,"methodId":129,"linkable":130,"proposed":77,"self":77},"FCGF","choy2019fcgf",true,{"name":132,"methodId":96,"linkable":77,"proposed":77,"self":77},"DIP",{"name":134,"methodId":96,"linkable":77,"proposed":77,"self":77},"Predator",{"name":136,"methodId":96,"linkable":77,"proposed":77,"self":77},"SpinNet",{"name":138,"methodId":96,"linkable":77,"proposed":77,"self":77},"D3Feat",{"name":140,"methodId":96,"linkable":77,"proposed":77,"self":77},"GeDi",{"name":142,"methodId":143,"linkable":130,"proposed":77,"self":77},"G-ICP","segal2009gicp",{"name":145,"methodId":146,"linkable":130,"proposed":77,"self":77},"STD, W = 1","std2023",{"name":148,"methodId":146,"linkable":130,"proposed":77,"self":77},"STD, W = 3",{"name":150,"methodId":146,"linkable":130,"proposed":77,"self":77},"STD, W = 5",{"name":152,"methodId":96,"linkable":77,"proposed":77,"self":77},"MapClosures, W = 1",{"name":154,"methodId":96,"linkable":77,"proposed":77,"self":77},"MapClosures, W = 3",{"name":156,"methodId":96,"linkable":77,"proposed":77,"self":77},"MapClosures, W = 5",{"name":158,"methodId":159,"linkable":130,"proposed":77,"self":77},"FPFH + FGR","zhou2016fgr",{"name":161,"methodId":162,"linkable":130,"proposed":77,"self":77},"FPFH + TEASER++","yang2021teaser",{"name":164,"methodId":96,"linkable":77,"proposed":77,"self":77},"FPFH + Quatro",{"name":166,"methodId":5,"linkable":130,"proposed":130,"self":130},"Proposed",{"name":168,"methodId":96,"linkable":77,"proposed":77,"self":77},"FPFH + FGR + G-ICP",{"name":170,"methodId":96,"linkable":77,"proposed":77,"self":77},"FPFH + TEASER + G-ICP",{"name":172,"methodId":96,"linkable":77,"proposed":77,"self":77},"FPFH + Quatro + G-ICP",{"name":174,"methodId":5,"linkable":130,"proposed":130,"self":130},"Proposed + G-ICP",[176,180,183,186,188,190,192,194,196,198,201,203,204,206,208,210,213,214,215,217,219,220,223,225,227,230,232,234,237,238,240,242,244,246,248,250,252,255,257,259,262,263,265,268,269,271,274,276,278,281,283,284,287,288,290,293,295,297,300,302,303,305,307,308,310,311],[177,177,177,178,179,177,179,179,177],0,25.9,-1,[177,181,177,182,179,177,179,179,177],1,0.57,[177,184,177,185,179,177,179,179,177],2,95.97,[181,177,177,187,179,177,179,179,177],6.47,[181,181,177,189,179,177,179,179,177],0.23,[181,184,177,191,179,177,179,179,177],99.82,[184,177,177,193,179,177,179,179,177],8.69,[184,181,177,195,179,177,179,179,177],0.44,[184,184,177,197,179,177,179,179,177],97.3,[199,177,177,200,179,177,179,179,177],3,5.6,[199,181,177,202,179,177,179,179,177],0.24,[199,184,177,191,179,177,179,179,177],[102,177,177,205,179,177,179,179,177],9.88,[102,181,177,207,179,177,179,179,177],0.47,[102,184,177,209,179,177,179,179,177],99.1,[211,177,177,212,179,177,179,179,177],5,11,[211,181,177,202,179,177,179,179,177],[211,184,177,191,179,177,179,179,177],[108,177,177,216,179,177,179,179,177],7.55,[108,181,177,218,179,177,179,179,177],0.33,[108,184,177,191,179,177,179,179,177],[221,177,177,222,179,177,179,179,177],7,8.56,[221,181,177,224,179,177,179,179,177],0.22,[221,184,177,226,179,177,179,179,177],37.95,[228,177,177,229,179,177,179,179,177],8,26.09,[228,181,177,231,179,177,179,179,177],0.69,[228,184,177,233,179,177,179,179,177],19.6,[235,177,177,236,179,177,179,179,177],9,20.94,[235,181,177,182,179,177,179,179,177],[235,184,177,239,179,177,179,179,177],30.58,[101,177,177,241,179,177,179,179,177],23.97,[101,181,177,243,179,177,179,179,177],0.66,[101,184,177,245,179,177,179,179,177],33.09,[212,177,177,247,179,177,179,179,177],38.5,[212,181,177,249,179,177,179,179,177],1.09,[212,184,177,251,179,177,179,179,177],69.38,[253,177,177,254,179,177,179,179,177],12,31.5,[253,181,177,256,179,177,179,179,177],0.95,[253,184,177,258,179,177,179,179,177],82.08,[260,177,177,261,179,177,179,179,177],13,32.27,[260,181,177,181,179,177,179,179,177],[260,184,177,264,179,177,179,179,177],86.64,[266,177,177,267,179,177,179,179,177],14,6.94,[266,181,177,218,179,177,179,179,177],[266,184,177,270,179,177,179,179,177],98.92,[272,177,177,273,179,177,179,179,177],15,9.36,[272,181,177,275,179,177,179,179,177],0.59,[272,184,177,277,179,177,179,179,177],99.64,[279,177,177,280,179,177,179,179,177],16,13.15,[279,181,177,282,179,177,179,179,177],0.94,[279,184,177,277,179,177,179,179,177],[285,177,177,286,179,177,179,179,177],17,18.1,[285,181,177,282,179,177,179,179,177],[285,184,177,289,179,177,179,179,177],100,[291,177,177,292,179,177,179,179,177],18,1.22,[291,181,177,294,179,177,179,179,177],0.04,[291,184,177,296,179,177,179,179,177],99.28,[298,177,177,299,179,177,179,179,177],19,1.1,[298,181,177,301,179,177,179,179,177],0.02,[298,184,177,277,179,177,179,179,177],[304,177,177,299,179,177,179,179,177],20,[304,181,177,306,179,177,179,179,177],0.03,[304,184,177,277,179,177,179,179,177],[309,177,177,299,179,177,179,179,177],21,[309,181,177,301,179,177,179,179,177],[309,184,177,289,179,177,179,179,177],[],[107],[],[],[317],"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",{"slug":319,"group":320,"sourceId":5,"sourceLabel":6,"table":321,"selfRows":184,"metrics":322,"seqs":328,"entrants":333,"cells":336,"outcomes":341,"locators":342,"hardware":344,"wordings":346,"notes":347},"lim2025kissmatcher-text-sec-iii-c","lim2025kissmatcher:Text Sec. III-C","Text Sec. III-C",[323,326],{"label":324,"unit":325,"statistic":18,"alignment":41},"speed improved approximately 4.5 times (single-threaded)","x",{"label":327,"unit":325,"statistic":18,"alignment":41},"speed improved approximately 2.4 times (multi-threaded)",[329],{"dataset":330,"sequence":331,"environment":332},"KITTI and MulRan","2-12 m loop closing test","vehicle",[334],{"name":335,"methodId":5,"linkable":130,"proposed":130,"self":130},"Faster-PFH vs FPFH",[337,339],[177,177,177,338,179,177,177,179,177],4.5,[177,181,177,340,179,177,177,179,177],2.4,[],[343],"Sec. III-C; Fig. 8(a) caption",[345],"not stated for these ratios (Fig. 8 caption names no CPU; Sec. III-C and Fig. 7 name an Intel Core i9-13900)",[],[348],"speed-up of Faster-PFH over FPFH while maintaining performance",{"slug":350,"group":351,"sourceId":5,"sourceLabel":6,"table":352,"selfRows":181,"metrics":353,"seqs":356,"entrants":361,"cells":364,"outcomes":366,"locators":368,"hardware":370,"wordings":371,"notes":372},"lim2025kissmatcher-text-sec-iv-e","lim2025kissmatcher:Text Sec. IV-E","Text Sec. IV-E",[354],{"label":355,"unit":325,"statistic":18,"alignment":41},"more than a 20x speed improvement",[357],{"dataset":358,"sequence":359,"environment":360},"not named (large-scale registration at kilometre level, Fig. 1(b))","voxelized clouds above 200K points","not stated",[362],{"name":363,"methodId":5,"linkable":130,"proposed":130,"self":130},"KISS-Matcher vs TEASER++ pipeline",[365],[177,177,177,304,177,177,179,179,177],[367],"lower bound (more than 20x)",[369],"Sec. IV-E; Fig. 1(b)",[],[],[373],"speed-up of the entire pipeline over the full TEASER++ pipeline in large-scale registration at the kilometre level (Sec. IV-E, Fig. 1(b)); the dataset behind Fig. 1(b) is not named",{"slug":375,"group":376,"sourceId":5,"sourceLabel":6,"table":377,"selfRows":181,"metrics":378,"seqs":382,"entrants":384,"cells":386,"outcomes":388,"locators":389,"hardware":391,"wordings":393,"notes":394},"lim2025kissmatcher-text-table-i-caption","lim2025kissmatcher:Text Table I caption","Text Table I caption",[379],{"label":380,"unit":381,"statistic":18,"alignment":41},"operates around 14 Hz for the entire pipeline","Hz",[383],{"dataset":86,"sequence":122,"environment":123},[385],{"name":7,"methodId":5,"linkable":130,"proposed":130,"self":130},[387],[177,177,177,266,179,177,177,179,177],[],[390],"Table I caption",[392],"Intel Core i9-13900 CPU",[],[395],"entire pipeline rate (feature extraction and matching to pose estimation) on the KITTI 10 m benchmark, stated in the Table I caption",[],1790510662913]