[{"data":1,"prerenderedAt":346},["ShallowReactive",2],{"method-imlsslam2018":3},{"method":4,"reference":57,"equipment":78,"figures":110,"results":111},{"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":27,"sensors":33,"platform":35,"estimator":37,"association":38,"timeModel":39,"deskew":40,"loopClosure":41,"globalOptimization":42,"mapRepresentation":43,"prior":42,"outputGeometry":44,"compute":45,"codeUrl":46,"codeLicense":47,"relatedVersions":48},"imlsslam2018","Deschaud, 2018","IMLS-SLAM","IMLS-SLAM: Scan-to-Model Matching Based on 3D Data",2018,"recent","C01","odometry_with_local_mapping","IMLS-SLAM 只使用 3D 旋轉式 LiDAR，以掃描對模型（scan-to-model）匹配估計位姿。模型是最近 n 個已定位掃描累積而成的點雲，並以隱式移動最小平方（IMLS）曲面表示；每次迭代先把取樣點投影到該曲面，再以線性化的點到平面最小平方求剛體轉換。取樣時依車體座標軸計算九組可觀測性分數，每組各取 s 個點，使旋轉與平移都受到約束；匹配前另以地面偵測與分群刪除尺寸小於門檻的物體，近似處理動態物體。系統沒有迴圈閉合，目前實作也不是即時運算。","LiDAR-only scan-to-model odometry that registers observability-sampled points to an implicit moving least squares surface built from the last n localized sweeps, with constant-velocity de-skewing and size-based small-object removal; no loop closure and not real time.","full_text_reviewed","peer_reviewed_published","supplementary","未在營建場域驗證；實驗為巴黎與里爾的市區車載資料以及 KITTI。其掃描對隱式曲面模型的配準，以及以尺寸門檻剔除小物體的做法，和工地中機具、人員造成的點雲污染問題相關；但以尺寸刪除物體也可能刪去工地上小型而靜態的構件（推論）。後續 LiDAR 里程計論文常以它作為 KITTI 比較基準，例如 KISS-ICP、CT-ICP、MULLS、SuMa++、GenZ-ICP 與 MOLA-LO 的比較表 [kissicp2023][cticp2022][mulls2021][sumapp2019][genzicp2025][molalo2025]。",[20,21],"public_benchmark","controlled_experiment",[23,24,25,26],"0.40% drift (16 m) over a 4 km, two-loop drive in Paris with a vertical HDL32, without loop closure (Sec. VI.A; Fig. 1).","KITTI training average 0.55% translational drift and 0.0015 deg\u002Fm; test set 0.69% and 0.0018 deg\u002Fm; lower drift than LOAM's published values on every training sequence (Table I; Sec. VI.B).","Ablations favour the proposed sampling (0.55% versus 0.64% random and 0.57% geometric stable sampling), a 100-scan model (versus 1.41% for scan-to-scan, n = 1) and object removal (0.55% versus 0.58%) (Tables II-IV).","Only about 7% of a scan (900 samples) is used for matching, so matching iterations take 0.05 s per scan (Sec. IV; Sec. VI.C).",[28,29,30,31,32],"Not real time: 1.25 s per scan on KITTI, dominated by normal computation on the 3D cloud and per-scan k-d tree rebuilding (Sec. VI.C).","KITTI drift is worse than on the HDL32 data; the author attributes this to intrinsic calibration distortion (a 0.22 deg vertical-angle correction was applied), GPS ground-truth errors above 5 m at the start of sequence 8, and more varied environments (Sec. VI.B).","LOAM's later KITTI website result (0.64%) was slightly better than IMLS-SLAM's test score (Sec. VI.B).","Dynamic handling removes every object whose bounding box is below 14 m x 14 m x 4 m, which is size-based rather than true motion detection (Sec. III).","The Lille run with the HDL32 tilted 60 degrees is evaluated only qualitatively (Sec. VI.A).",[34],"3D spinning LiDAR only (Velodyne HDL32 and HDL64 in the experiments); no IMU, GPS or camera",[36],"vehicle","iterative scan-to-model registration: each selected sample is projected onto the IMLS surface, then the rigid transform is found by linearized point-to-plane least squares under a small-angle assumption; fixed 20 iterations per scan (Sec. V; Sec. VI)","closest point in the model cloud (FLANN k-d tree) within radius r = 0.20 m; samples chosen from nine lists ranked by their contribution to observability of roll, pitch, yaw and the three translations (planarity-weighted), s = 100 per list (Sec. IV; Sec. VI)","one discrete pose per sweep end; poses inside a sweep linearly interpolated between the previous and current end pose, current pose predicted by constant relative motion (Sec. III)","linear interpolation between the previous end pose and the predicted end pose before matching, recomputed with the final pose after matching (Sec. III; Sec. V); not applied on KITTI, whose scans are already de-skewed (Sec. VI.B)","none (drift reported without any loop closure)","none","point cloud of the last n = 100 localized scans with normals, used as an implicit moving least squares (IMLS) surface (h = 0.06 m); the oldest scan is dropped as a new one is added and the k-d tree is rebuilt per scan (Sec. V; Sec. VI.C)","trajectory and accumulated de-skewed point cloud (Figs. 4-6)","C++ with FLANN and Eigen on one CPU core at 4 GHz, less than 1 GB RAM; not real time: 1.25 s per scan on KITTI (0.2 s normals, 1 s k-d tree, 0.05 s matching) (Sec. VI; Sec. VI.C)",null,"not_applicable (the paper links no code and no public implementation was located)",[49,53],{"relation":50,"title":51,"doi_or_url":52},"preprint","IMLS-SLAM: scan-to-model matching based on 3D data (arXiv v1)","https:\u002F\u002Farxiv.org\u002Fabs\u002F1802.08633",{"relation":54,"title":55,"doi_or_url":56},"repository_copy","HAL hal-01959570 (submitted version)","https:\u002F\u002Fhal.science\u002Fhal-01959570",{"id":5,"kind":58,"shortName":7,"title":8,"authors":59,"year":9,"venue":61,"venueType":62,"publisher":63,"volumeIssuePages":64,"doi":65,"arxivId":66,"url":67,"firstPublicDate":68,"publicationStatus":16,"metadataStatus":69,"fulltextStatus":15,"era":10,"classicReason":70,"codeUrl":46,"cluster":11,"topics":71,"mdpi":73,"verification":74,"label":6,"fulltextRoute":75,"versionRead":76,"addedByCensus":77},"method",[60],"Jean-Emmanuel Deschaud","2018 IEEE International Conference on Robotics and Automation (ICRA), Brisbane, Australia","conference","IEEE","pp. 2480-2485","10.1109\u002Ficra.2018.8460653","1802.08633","https:\u002F\u002Fdoi.org\u002F10.1109\u002FICRA.2018.8460653","2018-02-23","metadata_verified","not_applicable",[11,72],"C04",false,"confirmed","arXiv","arXiv v1 (2018-02-23), the only arXiv version; IEEE ICRA 2018 version of record not read",true,[79,86,91,98,104],{"category":80,"model":81,"canonical":82,"role":83,"dataset":46,"specs":84,"locator":85},"lidar","Velodyne HDL32","Velodyne HDL-32E","method input","32 laser beams, spinning at 10 Hz (100 ms per scan); mounted vertically on a car roof (Paris, 12951 scans) and tilted 60 degrees in pitch (Lille, 1500 scans)","Sec. VI; Sec. VI-A",{"category":87,"model":88,"canonical":88,"role":83,"dataset":46,"specs":89,"locator":90},"platform","car (vehicle roof mount, model not reported)","vehicle driven through Paris (two 2 km loops) and a square in Lille","Sec. VI-A; Fig. 1",{"category":80,"model":92,"canonical":93,"role":94,"dataset":95,"specs":96,"locator":97},"Velodyne HDL64","Velodyne HDL-64E","dataset sensor","KITTI odometry","vertical, on the roof of a car, 64 beams, 10 Hz; scans provided de-skewed","Sec. VI; Sec. VI-B",{"category":99,"model":100,"canonical":100,"role":101,"dataset":95,"specs":102,"locator":103},"gnss","GPS+IMU navigation system (model not reported)","reference or ground truth","ground truth for 11 training sequences; author notes errors above 5 m at the start of sequence 8","Sec. VI-B",{"category":105,"model":106,"canonical":106,"role":107,"dataset":46,"specs":108,"locator":109},"compute","single CPU core at 4 GHz (processor model not reported)","compute for runtime","one core, less than 1 GB RAM","Sec. VI",[],{"totalRows":112,"groupCount":113,"groups":114,"others":324},33,8,[115,218,252,282],{"slug":116,"group":117,"sourceId":5,"sourceLabel":6,"table":118,"selfRows":119,"metrics":120,"seqs":126,"entrants":153,"cells":159,"outcomes":212,"locators":213,"hardware":214,"wordings":215,"notes":216},"imlsslam2018-table-i","imlsslam2018:Table I","Table I",11,[121],{"label":122,"unit":123,"statistic":124,"alignment":125},"drift (%)","%","mean","not_reported",[127,130,133,136,139,141,143,145,147,149,151],{"dataset":95,"sequence":128,"environment":129},"00","Urban",{"dataset":95,"sequence":131,"environment":132},"01","Highway",{"dataset":95,"sequence":134,"environment":135},"02","Urban+Country",{"dataset":95,"sequence":137,"environment":138},"03","Country",{"dataset":95,"sequence":140,"environment":138},"04",{"dataset":95,"sequence":142,"environment":129},"05",{"dataset":95,"sequence":144,"environment":129},"06",{"dataset":95,"sequence":146,"environment":129},"07",{"dataset":95,"sequence":148,"environment":135},"08",{"dataset":95,"sequence":150,"environment":135},"09",{"dataset":95,"sequence":152,"environment":135},"10",[154,157],{"name":155,"methodId":156,"linkable":77,"proposed":73,"self":73},"LOAM [7] (results taken from paper)","loam2017_auro",{"name":158,"methodId":5,"linkable":77,"proposed":77,"self":77},"Our SLAM (IMLS-SLAM)",[160,164,167,169,171,174,176,179,181,184,186,189,191,194,195,198,199,201,203,206,208,211],[161,161,161,162,163,161,163,163,161],0,0.78,-1,[165,161,161,166,163,161,163,163,161],1,0.5,[161,161,165,168,163,161,163,163,161],1.43,[165,161,165,170,163,161,163,163,161],0.82,[161,161,172,173,163,161,163,163,161],2,0.92,[165,161,172,175,163,161,163,163,161],0.53,[161,161,177,178,163,161,163,163,161],3,0.86,[165,161,177,180,163,161,163,163,161],0.68,[161,161,182,183,163,161,163,163,161],4,0.71,[165,161,182,185,163,161,163,163,161],0.33,[161,161,187,188,163,161,163,163,161],5,0.57,[165,161,187,190,163,161,163,163,161],0.32,[161,161,192,193,163,161,163,163,161],6,0.65,[165,161,192,185,163,161,163,163,161],[161,161,196,197,163,161,163,163,161],7,0.63,[165,161,196,185,163,161,163,163,161],[161,161,113,200,163,161,163,163,161],1.12,[165,161,113,202,163,161,163,163,161],0.8,[161,161,204,205,163,161,163,163,161],9,0.77,[165,161,204,207,163,161,163,163,161],0.55,[161,161,209,210,163,161,163,163,161],10,0.79,[165,161,209,175,163,161,163,163,161],[],[118],[],[],[217],"KITTI odometry training sequences 00-10, HDL64; translation drift (%) with the KITTI metric; LOAM values copied from the LOAM Autonomous Robots paper [7]",{"slug":219,"group":220,"sourceId":5,"sourceLabel":6,"table":221,"selfRows":182,"metrics":222,"seqs":225,"entrants":229,"cells":238,"outcomes":246,"locators":247,"hardware":248,"wordings":249,"notes":250},"imlsslam2018-table-iv","imlsslam2018:Table IV","Table IV",[223],{"label":224,"unit":123,"statistic":124,"alignment":125},"Drift on KITTI training dataset",[226],{"dataset":95,"sequence":227,"environment":228},"training 00-10 (overall)","vehicle, mixed",[230,232,234,236],{"name":231,"methodId":5,"linkable":77,"proposed":77,"self":77},"IMLS-SLAM, n = 1 scan",{"name":233,"methodId":5,"linkable":77,"proposed":77,"self":77},"IMLS-SLAM, n = 5 scans",{"name":235,"methodId":5,"linkable":77,"proposed":77,"self":77},"IMLS-SLAM, n = 10 scans",{"name":237,"methodId":5,"linkable":77,"proposed":77,"self":77},"IMLS-SLAM, n = 100 scans",[239,241,243,245],[161,161,161,240,163,161,163,163,161],1.41,[165,161,161,242,163,161,163,163,161],0.58,[172,161,161,244,163,161,163,163,161],0.56,[177,161,161,207,163,161,163,163,161],[],[221],[],[],[251],"Ablation of number of scans n kept in the model; drift on the whole KITTI training dataset",{"slug":253,"group":254,"sourceId":5,"sourceLabel":6,"table":255,"selfRows":182,"metrics":256,"seqs":262,"entrants":266,"cells":268,"outcomes":276,"locators":277,"hardware":278,"wordings":279,"notes":280},"imlsslam2018-text-sec-vi-b","imlsslam2018:Text Sec. VI-B","Text Sec. VI-B",[257,259],{"label":258,"unit":123,"statistic":124,"alignment":125},"drift in translation",{"label":260,"unit":261,"statistic":124,"alignment":125},"error in rotation (deg\u002Fm)","deg\u002Fm",[263,264],{"dataset":95,"sequence":227,"environment":228},{"dataset":95,"sequence":265,"environment":228},"test 11-21 (KITTI website)",[267],{"name":7,"methodId":5,"linkable":77,"proposed":77,"self":77},[269,270,272,274],[161,161,161,207,163,161,163,163,161],[161,165,161,271,163,161,163,163,161],0.0015,[161,161,165,273,163,161,163,163,161],0.69,[161,165,165,275,163,161,163,163,161],0.0018,[],[103],[],[],[281],"KITTI odometry, overall values stated in text (training set with ground truth; test set from KITTI website)",{"slug":283,"group":284,"sourceId":5,"sourceLabel":6,"table":285,"selfRows":182,"metrics":286,"seqs":296,"entrants":299,"cells":308,"outcomes":316,"locators":317,"hardware":319,"wordings":321,"notes":322},"imlsslam2018-text-sec-vi-c","imlsslam2018:Text Sec. VI-C","Text Sec. VI-C",[287,290,292,294],{"label":288,"unit":289,"statistic":125,"alignment":70},"total time per scan","s",{"label":291,"unit":289,"statistic":125,"alignment":70},"normal computation per scan",{"label":293,"unit":289,"statistic":125,"alignment":70},"k-d tree construction per sweep (n = 100)",{"label":295,"unit":289,"statistic":125,"alignment":70},"matching iterations per scan",[297],{"dataset":95,"sequence":298,"environment":36},"all",[300,302,304,306],{"name":301,"methodId":5,"linkable":77,"proposed":77,"self":77},"IMLS SLAM",{"name":303,"methodId":5,"linkable":77,"proposed":77,"self":77},"IMLS SLAM (normal computation)",{"name":305,"methodId":5,"linkable":77,"proposed":77,"self":77},"IMLS SLAM (k-d tree rebuild)",{"name":307,"methodId":5,"linkable":77,"proposed":77,"self":77},"IMLS SLAM (matching)",[309,311,313,314],[161,161,161,310,163,161,161,163,161],1.25,[165,165,161,312,163,161,161,163,161],0.2,[172,172,161,165,163,161,161,163,161],[177,177,161,315,163,161,161,163,161],0.05,[],[318],"Sec. VI-C",[320],"one CPU core at 4 GHz, less than 1 GB RAM",[],[323],"Processing time per scan on KITTI (normals from 3D points since raw range images are unavailable)",[325,330,335,340],{"group":326,"slug":327,"sourceLabel":6,"table":328,"selfRows":177,"datasets":329},"imlsslam2018:Table III","imlsslam2018-table-iii","Table III",[95],{"group":331,"slug":332,"sourceLabel":6,"table":333,"selfRows":177,"datasets":334},"imlsslam2018:Table V","imlsslam2018-table-v","Table V",[95],{"group":336,"slug":337,"sourceLabel":6,"table":338,"selfRows":172,"datasets":339},"imlsslam2018:Table II","imlsslam2018-table-ii","Table II",[95],{"group":341,"slug":342,"sourceLabel":6,"table":343,"selfRows":172,"datasets":344},"imlsslam2018:Text Sec. VI-A","imlsslam2018-text-sec-vi-a","Text Sec. VI-A",[345],"own HDL32 Paris dataset",1790510662817]