[{"data":1,"prerenderedAt":528},["ShallowReactive",2],{"method-koide2021vgicp":3},{"method":4,"reference":53,"equipment":75,"figures":93,"results":94},{"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":24,"sensors":29,"platform":31,"estimator":34,"association":35,"timeModel":36,"deskew":37,"loopClosure":38,"globalOptimization":38,"mapRepresentation":39,"prior":40,"outputGeometry":41,"compute":42,"codeUrl":43,"codeLicense":44,"relatedVersions":45},"koide2021vgicp","Koide et al., 2021b","VGICP","Voxelized GICP for Fast and Accurate 3D Point Cloud Registration",2021,"recent","C02","registration_component","VGICP 延伸 GICP，以體素化取代耗時的最近鄰搜尋：每個體素彙整其內各點的分布（而非像 NDT 直接由點位置計算分布），形成分布對多分布的對應，即使體素內點數少也能得到有效分布。體素化使最佳化容易平行化，作者報告 CPU 約 30 Hz、GPU 約 120 Hz，精度與 GICP 相當且對體素解析度較不敏感。","VGICP voxelizes GICP by aggregating per-point covariances per voxel, avoiding nearest-neighbour search and enabling parallel CPU\u002FGPU registration with GICP-level accuracy.","full_text_reviewed","peer_reviewed_published","main_body","not_reported（實驗為模擬與約 120 m 的 HDL-32E 序列，場域類型未明示）",[20],"simulation",[22,23],"accuracy comparable to GICP but substantially faster (abstract)","robust to voxel resolution changes, unlike NDT (Sec. IV-B, Sec. V)",[25,26,27,28],"voxelization may affect convergence when the initial guess is far from the true pose (Sec. V, future work)","real-sequence reference built by aligning last frame to first with GICP, so reference is not independent (Sec. IV-B; independence concern is reviewer inference)","GPU brute-force nearest-neighbour search was slower than the CPU parallel KD-tree in this evaluation (Sec. IV-A)","(reviewer observation) evaluation is consecutive-frame odometry only; no loop closure, map-quality or engineering-accuracy evaluation (Sec. IV)",[30],"3D LiDAR (Velodyne HDL-32E real and simulated)",[32,33],"simulation (authors' ray-casting LiDAR simulator using Velodyne HDL-32e parameters)","real HDL-32e recordings, eight sequences of about 120 m; carrying platform not stated","GICP-style least squares with voxel-based distribution-to-multi-distribution residuals; implementation uses Gauss-Newton-type optimizer (Sec. III, IV-B)","per-point covariances from k nearest neighbours (for example k = 20) found with a KD-tree and regularised to eigenvalues (1, 1, epsilon); the target is voxelized by averaging point means and covariances per voxel; each source point is matched to the voxel it falls in (distribution-to-multi-distribution), so no nearest-neighbour search is needed during optimization","discrete poses (scan-to-scan odometry in experiments)","not_reported","none","voxel map of aggregated point distributions (Gaussian voxelmap)","initial guess required (Algorithm 1)","6-DoF rigid transformation","single-threaded GICP 189 ms and VGICP 156 ms versus PCL GICP 201 ms; multi-threaded GICP 68 ms and VGICP 50 ms; GPU VGICP 6 ms to optimize (simulated data, Sec. IV-A); on real data about 30 fps on CPU and 120 fps on GPU for about 15,000-point frames (Table II); Intel Core i9-9900K and NVIDIA GeForce RTX2080Ti (Sec. IV-A)","https:\u002F\u002Fgithub.com\u002Fkoide3\u002Ffast_gicp","BSD-3-Clause (LICENSE file checked)",[46,50],{"relation":47,"title":48,"doi_or_url":49},"preprint","EasyChair Preprint 2703 (dated 2020-02-18)","https:\u002F\u002Feasychair.org\u002Fpublications\u002Fpreprint\u002FftvV",{"relation":51,"title":52,"doi_or_url":43},"code_release","fast_gicp",{"id":5,"kind":54,"shortName":7,"title":8,"authors":55,"year":9,"venue":60,"venueType":61,"publisher":62,"volumeIssuePages":63,"doi":64,"arxivId":65,"url":66,"firstPublicDate":67,"publicationStatus":16,"metadataStatus":68,"fulltextStatus":15,"era":10,"classicReason":69,"codeUrl":43,"cluster":11,"topics":70,"mdpi":71,"verification":72,"label":6,"fulltextRoute":73,"versionRead":74,"addedByCensus":71},"method",[56,57,58,59],"Kenji Koide","Masashi Yokozuka","Shuji Oishi","Atsuhiko Banno","2021 IEEE International Conference on Robotics and Automation (ICRA)","conference","IEEE","pp. 11054-11059","10.1109\u002Ficra48506.2021.9560835",null,"https:\u002F\u002Fstaff.aist.go.jp\u002Fshuji.oishi\u002Fassets\u002Fpapers\u002Fpreprint\u002FVoxelGICP_ICRA2021.pdf","2020-02-18","metadata_verified","not_applicable",[11],false,"confirmed","author copy","author preprint hosted on the AIST staff page (6 pages, ICRA 2021 layout); not compared line by line with the IEEE Xplore version of record",[76,84,90],{"category":77,"model":78,"canonical":79,"role":80,"dataset":81,"specs":82,"locator":83},"lidar","Velodyne HDL-32e","Velodyne HDL-32E","method input","authors' real HDL-32e sequences","about 15,000 points per frame; eight sequences of about 120 m","Sec. IV-B",{"category":85,"model":86,"canonical":86,"role":87,"dataset":65,"specs":88,"locator":89},"compute","Intel Core i9-9900K","compute for runtime","CPU used for all methods","Sec. IV-A",{"category":85,"model":91,"canonical":91,"role":87,"dataset":65,"specs":92,"locator":89},"NVIDIA Geforce RTX2080Ti","GPU for VGICP GPU version",[],{"totalRows":95,"groupCount":96,"groups":97,"others":515},24,6,[98,193,429,469],{"slug":99,"group":100,"sourceId":5,"sourceLabel":6,"table":101,"selfRows":102,"metrics":103,"seqs":120,"entrants":125,"cells":147,"outcomes":186,"locators":187,"hardware":188,"wordings":190,"notes":191},"koide2021vgicp-table-ii","koide2021vgicp:Table II","Table II",8,[104,107,109,113,116,118],{"label":105,"unit":106,"statistic":37,"alignment":69},"FPS (CPU)","fps",{"label":108,"unit":106,"statistic":37,"alignment":69},"FPS (GPU)",{"label":110,"unit":111,"statistic":37,"alignment":112},"Translation [m] at last frame (0.852 ± 0.289)","m","first-pose",{"label":114,"unit":115,"statistic":37,"alignment":112},"Rotation [deg] at last frame (0.049 ± 0.013)","deg",{"label":117,"unit":111,"statistic":37,"alignment":112},"Translation [m] at last frame (1.177 ± 0.456)",{"label":119,"unit":115,"statistic":37,"alignment":112},"Rotation [deg] at last frame (0.048 ± 0.026)",[121],{"dataset":122,"sequence":123,"environment":124},"authors' HDL-32e sequences","8 sequences (aggregate)","real environment (Fig. 5; type not described)",[126,129,131,134,136,138,140,142,144],{"name":127,"methodId":5,"linkable":128,"proposed":128,"self":128},"VGICP (0.5m)",true,{"name":130,"methodId":5,"linkable":128,"proposed":128,"self":128},"VGICP (1.0m)",{"name":132,"methodId":133,"linkable":128,"proposed":71,"self":71},"GICP (ours)","segal2009gicp",{"name":135,"methodId":133,"linkable":128,"proposed":71,"self":71},"GICP (PCL)",{"name":137,"methodId":65,"linkable":71,"proposed":71,"self":71},"NDT (0.5m)",{"name":139,"methodId":65,"linkable":71,"proposed":71,"self":71},"NDT (1.0m)",{"name":141,"methodId":65,"linkable":71,"proposed":71,"self":71},"NDT (2.0m)",{"name":143,"methodId":65,"linkable":71,"proposed":71,"self":71},"NDT (4.0m)",{"name":145,"methodId":146,"linkable":128,"proposed":71,"self":71},"ICP","besl1992icp",[148,152,155,158,161,163,165,168,171,173,175,177,179,181,184],[149,149,149,150,151,149,149,151,149],0,28.7,-1,[149,153,149,154,151,149,149,151,149],1,116.5,[149,156,149,157,151,149,151,151,149],2,0.852,[149,159,149,160,151,149,151,151,149],3,0.049,[153,149,149,162,151,149,149,151,149],30.4,[153,153,149,164,151,149,149,151,149],120.1,[153,166,149,167,151,149,151,151,149],4,1.177,[153,169,149,170,151,149,151,151,149],5,0.048,[156,149,149,172,151,149,149,151,149],20.5,[159,149,149,174,151,149,149,151,149],5.2,[166,149,149,176,151,149,149,151,149],8.3,[169,149,149,178,151,149,149,151,149],10.3,[96,149,149,180,151,149,149,151,149],10.4,[182,149,149,183,151,149,149,151,149],7,9.3,[102,149,149,185,151,149,149,151,149],11,[],[101],[189],"not restated in Sec. IV-B; Sec. IV-A states all methods were run on an Intel Core i9-9900K and NVIDIA GeForce RTX2080Ti",[],[192],"eight real Velodyne HDL-32e sequences of about 120 m, about 15,000 points per frame; consecutive-frame registration; error of the last frame against a reference obtained by aligning the last frame to the first with GICP; the ± term is not defined in the paper (presumably spread over the eight sequences)",{"slug":194,"group":195,"sourceId":196,"sourceLabel":197,"table":198,"selfRows":96,"metrics":199,"seqs":206,"entrants":215,"cells":251,"outcomes":422,"locators":424,"hardware":425,"wordings":426,"notes":427},"lim2024quatropp-table-6","lim2024quatropp:Table 6","lim2024quatropp","Lim et al., 2024","Table 6",[200,203],{"label":201,"unit":202,"statistic":37,"alignment":37},"trel","%",{"label":204,"unit":205,"statistic":37,"alignment":37},"rrel","deg\u002F100m",[207,211,213],{"dataset":208,"sequence":209,"environment":210},"KITTI","Seq. 00, Delta = 1","vehicle, urban driving (Velodyne HDL-64E)",{"dataset":208,"sequence":212,"environment":210},"Seq. 00, Delta = 3",{"dataset":208,"sequence":214,"environment":210},"Seq. 00, Delta = 5",[216,217,219,220,223,226,228,230,233,235,237,239,241,244,247,249],{"name":145,"methodId":146,"linkable":128,"proposed":71,"self":71},{"name":218,"methodId":133,"linkable":128,"proposed":71,"self":71},"G-ICP",{"name":7,"methodId":5,"linkable":128,"proposed":71,"self":128},{"name":221,"methodId":222,"linkable":128,"proposed":71,"self":71},"FGR","zhou2016fgr",{"name":224,"methodId":225,"linkable":128,"proposed":71,"self":71},"TEASER++","yang2021teaser",{"name":227,"methodId":65,"linkable":71,"proposed":128,"self":71},"Quatro (Ours)",{"name":229,"methodId":196,"linkable":128,"proposed":128,"self":71},"Quatro++ (Ours)",{"name":231,"methodId":232,"linkable":128,"proposed":71,"self":71},"LO-Net","lonet2019",{"name":234,"methodId":232,"linkable":128,"proposed":71,"self":71},"LO-Net+M",{"name":236,"methodId":65,"linkable":71,"proposed":71,"self":71},"DMLO†",{"name":238,"methodId":65,"linkable":71,"proposed":71,"self":71},"DMLO+M†",{"name":240,"methodId":65,"linkable":71,"proposed":71,"self":71},"A-LOAM + StickyPillars†",{"name":242,"methodId":243,"linkable":128,"proposed":71,"self":71},"SuMa","suma2018",{"name":245,"methodId":246,"linkable":128,"proposed":71,"self":71},"A-LOAM","aloam_software",{"name":248,"methodId":65,"linkable":71,"proposed":128,"self":71},"Quatro-c2f (Ours)",{"name":250,"methodId":196,"linkable":128,"proposed":128,"self":71},"Quatro++-c2f (Ours)",[252,254,256,258,260,262,264,266,268,270,272,274,276,278,280,282,284,286,288,290,292,294,296,298,300,302,304,306,308,310,311,312,314,316,318,320,322,324,326,327,329,331,333,335,337,338,339,340,341,343,345,346,347,348,349,352,354,355,356,357,358,361,362,363,364,365,366,368,370,372,374,376,378,381,383,385,387,389,391,394,396,398,400,402,404,406,408,410,411,412,413,415,416,417,418,420],[149,149,149,253,151,149,151,151,149],6.88,[149,153,149,255,151,149,151,151,149],2.99,[149,149,153,257,151,149,151,151,149],21.92,[149,153,153,259,151,149,151,151,149],8.7,[149,149,156,261,151,149,151,151,149],21.14,[149,153,156,263,151,149,151,151,149],8.51,[153,149,149,265,151,149,151,151,149],1.26,[153,153,149,267,151,149,151,151,149],0.45,[153,149,153,269,151,149,151,151,149],5.5,[153,153,153,271,151,149,151,151,149],1.45,[153,149,156,273,151,149,151,151,149],14.2,[153,153,156,275,151,149,151,151,149],3.32,[156,149,149,277,151,149,151,151,149],1.03,[156,153,149,279,151,149,151,151,149],0.3,[156,149,153,281,151,149,151,151,149],11.83,[156,153,153,283,151,149,151,151,149],1.65,[156,149,156,285,151,149,151,151,149],19.11,[156,153,156,287,151,149,151,151,149],6.32,[159,149,149,289,151,149,151,151,149],2.73,[159,153,149,291,151,149,151,151,149],0.69,[159,149,153,293,151,149,151,151,149],7.17,[159,153,153,295,151,149,151,151,149],1.58,[159,149,156,297,151,149,151,151,149],14.66,[159,153,156,299,151,149,151,151,149],4.12,[166,149,149,301,151,149,151,151,149],2.11,[166,153,149,303,151,149,151,151,149],0.91,[166,149,153,305,151,149,151,151,149],2.64,[166,153,153,307,151,149,151,151,149],1.11,[166,149,156,309,151,149,151,151,149],3.19,[166,153,156,303,151,149,151,151,149],[169,149,149,271,151,149,151,151,149],[169,153,149,313,151,149,151,151,149],0.41,[169,149,153,315,151,149,151,151,149],1.38,[169,153,153,317,151,149,151,151,149],0.24,[169,149,156,319,151,149,151,151,149],1.94,[169,153,156,321,151,149,151,151,149],0.46,[96,149,149,323,151,149,151,151,149],1.9,[96,153,149,325,151,149,151,151,149],0.53,[96,149,153,271,151,149,151,151,149],[96,153,153,328,151,149,151,151,149],0.32,[96,149,156,330,151,149,151,151,149],0.99,[96,153,156,332,151,149,151,151,149],0.28,[182,149,149,334,151,149,151,151,149],1.47,[182,153,149,336,151,149,151,151,149],0.72,[182,149,153,65,149,149,151,151,149],[182,153,153,65,149,149,151,151,149],[182,149,156,65,149,149,151,151,149],[182,153,156,65,149,149,151,151,149],[102,149,149,342,151,149,151,151,149],0.78,[102,153,149,344,151,149,151,151,149],0.42,[102,149,153,65,149,149,151,151,149],[102,153,153,65,149,149,151,151,149],[102,149,156,65,149,149,151,151,149],[102,153,156,65,149,149,151,151,149],[350,149,149,351,151,149,151,151,149],9,0.83,[350,153,149,353,151,149,151,151,149],0.44,[350,149,153,65,149,149,151,151,149],[350,153,153,65,149,149,151,151,149],[350,149,156,65,149,149,151,151,149],[350,153,156,65,149,149,151,151,149],[359,149,149,360,151,149,151,151,149],10,0.73,[359,153,149,353,151,149,151,151,149],[359,149,153,65,149,149,151,151,149],[359,153,153,65,149,149,151,151,149],[359,149,156,65,149,149,151,151,149],[359,153,156,65,149,149,151,151,149],[185,149,149,367,151,149,151,151,149],0.65,[185,153,149,369,151,149,151,151,149],0.26,[185,149,153,371,151,149,151,151,149],0.79,[185,153,153,373,151,149,151,151,149],0.31,[185,149,156,375,151,149,151,151,149],1.29,[185,153,156,377,151,149,151,151,149],0.48,[379,149,149,380,151,149,151,151,149],12,0.68,[379,153,149,382,151,149,151,151,149],0.23,[379,149,153,384,151,149,151,151,149],1.69,[379,153,153,386,151,149,151,151,149],0.61,[379,149,156,388,151,149,151,151,149],2.36,[379,153,156,390,151,149,151,151,149],0.51,[392,149,149,393,151,149,151,151,149],13,0.7,[392,153,149,395,151,149,151,151,149],0.27,[392,149,153,397,151,149,151,151,149],0.97,[392,153,153,399,151,149,151,151,149],0.38,[392,149,156,401,151,149,151,151,149],31.16,[392,153,156,403,151,149,151,151,149],12.1,[405,149,149,367,151,149,151,151,149],14,[405,153,149,407,151,149,151,151,149],0.21,[405,149,153,409,151,149,151,151,149],0.67,[405,153,153,407,151,149,151,151,149],[405,149,156,409,151,149,151,151,149],[405,153,156,407,151,149,151,151,149],[414,149,149,380,151,149,151,151,149],15,[414,153,149,382,151,149,151,151,149],[414,149,153,386,151,149,151,151,149],[414,153,153,407,151,149,151,151,149],[414,149,156,419,151,149,151,151,149],0.5,[414,153,156,421,151,149,151,151,149],0.2,[423],"not_reported (N\u002FA in table; not available from the original paper)",[198],[],[],[428],"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":430,"group":431,"sourceId":432,"sourceLabel":433,"table":434,"selfRows":166,"metrics":435,"seqs":445,"entrants":450,"cells":453,"outcomes":462,"locators":463,"hardware":465,"wordings":466,"notes":467},"koide2024smallgicp-text-benchmark-md-accuracy","koide2024smallgicp:Text BENCHMARK.md Accuracy","koide2024smallgicp","Koide, 2024","Text BENCHMARK.md Accuracy",[436,439,441,443],{"label":437,"unit":438,"statistic":37,"alignment":37},"APE = 6.791 +- 3.215","not_stated",{"label":440,"unit":438,"statistic":37,"alignment":37},"RPE(100) = 1.253 +- 0.734",{"label":442,"unit":438,"statistic":37,"alignment":37},"RPE(400) = 6.315 +- 3.011",{"label":444,"unit":438,"statistic":37,"alignment":37},"RPE(800) = 10.367 +- 6.147",[446],{"dataset":447,"sequence":448,"environment":449},"KITTI odometry sequence 00","00","vehicle, urban driving",[451],{"name":452,"methodId":5,"linkable":128,"proposed":71,"self":128},"fast_vgicp",[454,456,458,460],[149,149,149,455,151,149,151,151,149],6.791,[149,153,149,457,151,149,151,151,149],1.253,[149,156,149,459,151,149,151,151,149],6.315,[149,159,149,461,151,149,151,151,149],10.367,[],[464],"BENCHMARK.md, Accuracy",[],[],[468],"repository documentation linked from the paper (BENCHMARK.md, master branch, fetched 2026-09-25), not peer-reviewed text; odometry benchmark on KITTI 00; units and the meaning of '+-' and of the RPE window (100, 400, 800) are not stated; value is the number before '+-'",{"slug":470,"group":471,"sourceId":5,"sourceLabel":6,"table":472,"selfRows":159,"metrics":473,"seqs":480,"entrants":484,"cells":496,"outcomes":508,"locators":509,"hardware":510,"wordings":512,"notes":513},"koide2021vgicp-text-sec-iv-a","koide2021vgicp:Text Sec. IV-A","Text Sec. IV-A",[474,478],{"label":475,"unit":476,"statistic":477,"alignment":69},"processing time","ms","mean",{"label":479,"unit":476,"statistic":477,"alignment":69},"time to optimize on GPU",[481],{"dataset":482,"sequence":483,"environment":20},"authors' simulated LiDAR sequence","simulated sequence",[485,487,489,490,492,494],{"name":486,"methodId":133,"linkable":128,"proposed":71,"self":71},"GICP (ours, single-thread)",{"name":488,"methodId":5,"linkable":128,"proposed":128,"self":128},"VGICP (single-thread)",{"name":135,"methodId":133,"linkable":128,"proposed":71,"self":71},{"name":491,"methodId":133,"linkable":128,"proposed":71,"self":71},"GICP (ours, multi-thread)",{"name":493,"methodId":5,"linkable":128,"proposed":128,"self":128},"VGICP (multi-thread)",{"name":495,"methodId":5,"linkable":128,"proposed":128,"self":128},"VGICP (GPU)",[497,499,501,503,505,507],[149,149,149,498,151,149,149,151,149],189,[153,149,149,500,151,149,149,151,149],156,[156,149,149,502,151,149,149,151,149],201,[159,149,149,504,151,149,149,151,149],68,[166,149,149,506,151,149,149,151,149],50,[169,153,149,96,151,149,149,151,149],[],[89],[511],"Intel Core i9-9900K CPU; NVIDIA GeForce RTX2080Ti GPU",[],[514],"average processing time per scan on the simulated sequence, as stated in the Sec. IV-A text (Fig. 4 is a plot and was not read off); the text does not say explicitly whether each stated time includes the covariance preprocessing; the GPU figure is described as the time to optimize",[516,521],{"group":517,"slug":518,"sourceLabel":6,"table":519,"selfRows":156,"datasets":520},"koide2021vgicp:Table I","koide2021vgicp-table-i","Table I",[482],{"group":522,"slug":523,"sourceLabel":524,"table":525,"selfRows":153,"datasets":526},"genzicp2025:Table III","genzicp2025-table-iii","Lee et al., 2025a","Table III",[527],"KITTI odometry",1790510664463]