[{"data":1,"prerenderedAt":481},["ShallowReactive",2],{"method-schauer2018peopleremover":3},{"method":4,"reference":56,"equipment":75,"figures":83,"results":84},{"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":24,"limitations":30,"sensors":36,"platform":41,"estimator":45,"association":46,"timeModel":47,"deskew":47,"loopClosure":47,"globalOptimization":48,"mapRepresentation":49,"prior":50,"outputGeometry":51,"compute":52,"codeUrl":53,"codeLicense":54,"relatedVersions":55},"schauer2018peopleremover","Schauer & Nuchter, 2018","Peopleremover","The Peopleremover, Removing Dynamic Objects From 3-D Point Cloud Data by Traversing a Voxel Occupancy Grid",2018,"recent","C13","map_representation_or_reconstruction","本法以已配準的多站或多切片點雲建立全域體素網格，每個體素只記錄有哪些掃描在其中量到點。從每個感測器原點沿視線走訪到各量測點，若某體素被其他掃描看穿為空，體素內的點即判定為動態並移除；為避免斜掃表面與取樣不均造成誤判，以最近點的「陰影」與法向量限制走訪距離，再以叢集過濾孤立誤判，並可依掃描編號做次體素移除。演算法對掃描數與點數呈線性複雜度，只有體素大小一個參數。","Ray traversal through a voxel occupancy grid over registered scans to detect occupancy differences and remove dynamic objects.","full_text_reviewed","peer_reviewed_published","background","未在工地測試。引言把開挖工地（excavation site）與工廠產線列為不宜為掃描而停工的場景，也提到礦場進度監測；作者先前也把方法用於汽車產線的移動測繪。對需要在施工持續進行時掃描的工地，本法可作為配準後移除人員與機具的後處理步驟（推論）。",[20,21,22,23],"simulation","public_benchmark","completed_building","independent_reference",[25,26,27,28,29],"Works for mobile-mapping scan slices and terrestrial scans (abstract)","F1 equal to Underwood et al. on sim (0.98) and lecturehall (0.96) and higher on carpark (0.83 vs 0.78) (Table I)","Campus dataset processed in 13.1 h versus 12.8 days for pairwise comparison with Underwood et al. (Table I)","Only one parameter (voxel size); a 17.5 cm voxel on lecturehall cut time by 18% to 567 s with F1 0.95 (Sec. IX)","Conservative: volumes seen by a single scan are left untouched, so occlusion is not treated as change (Sec. II, III)",[31,32,33,34,35],"Runtime per iteration reported as far slower than ERASOR on SemanticKITTI (Lim et al. 2021, Table III) (cross-study)","Lower F1 than Underwood et al. on the noisy lab dataset with very small dynamic objects (0.42 vs 0.71) (Table I, Sec. IX)","False positives from registration errors, sensor noise, wrong normals, mirrors and transparent objects, and from the sub-voxel option (Sec. IX)","False negatives where a volume is seen by one scan only, is shadowed by near points (scanner on the ground), the voxel is too small or ranges are far, or objects are smaller than the minimum cluster size (Sec. IX)","Best voxel size differs per dataset and currently needs labelled training data (Sec. IX)",[37,38,39,40],"3D laser scans only: Riegl VZ-400 TLS for the authors' lecturehall, campus and wrzburg datasets","third-party datasets of Underwood et al. (sim, lab, carpark; sensor not stated in this paper)","mobile-mapping scan slices from an automotive production line (authors' earlier work)","stated as compatible in principle with RADAR, RGB-D or stereo point clouds (not tested)",[42,43,44],"static terrestrial laser scanner (tripod; one test with the scanner placed on the ground)","mobile mapping scan slices (automotive production line dataset)","simulation (sim dataset)","not_applicable (post-registration cleaning)","Amanatides-Woo voxel traversal (made stricter and free of floating-point accumulation) from each sensor origin to each point; traversal aborts at voxels holding the same scan identifier; search distances clipped by point 'shadows' using normals and per-scan sphere quadtrees; small clusters of free voxels reset to static; optional sub-voxel removal by scan identifier","not_applicable","none","global regular voxel grid storing only the set of scan identifiers per voxel (no point coordinates); single parameter voxel size (0.1 to 0.6 m in Table I); C++ standard library containers","registered scans with known sensor origins (wrzburg registered with slam6D from 3DTK); each volume to be cleaned must be observed by at least two scans taken far enough apart in time","cleaned point cloud without dynamic objects","offline; linear in the number of scans and points; single-threaded benchmarks although traversal parallelizes; e.g. 687 s for lecturehall (44.6 million points, 2 scans) and 13.1 h for campus (2.2 billion points, 146 scans); hardware not reported; whole point cloud held in memory",null,"not_verified",[],{"id":5,"kind":57,"shortName":7,"title":58,"authors":59,"year":9,"venue":62,"venueType":63,"publisher":64,"volumeIssuePages":65,"doi":66,"arxivId":53,"url":67,"firstPublicDate":68,"publicationStatus":16,"metadataStatus":69,"fulltextStatus":15,"era":10,"classicReason":47,"codeUrl":53,"cluster":11,"topics":70,"mdpi":71,"verification":72,"label":6,"fulltextRoute":73,"versionRead":74,"addedByCensus":71},"component","The Peopleremover—Removing Dynamic Objects From 3-D Point Cloud Data by Traversing a Voxel Occupancy Grid",[60,61],"Johannes Schauer","Andreas Nuchter","IEEE Robotics and Automation Letters","journal","IEEE","3(3), pp. 1679-1686","10.1109\u002Flra.2018.2801797","https:\u002F\u002Fdoi.org\u002F10.1109\u002FLRA.2018.2801797","2018-02-05","metadata_verified",[11],false,"confirmed","NTU institutional (curl)","Version of record, IEEE Robotics and Automation Letters 3(3):1679-1686, July 2018 (IEEE Xplore PDF)",[76],{"category":77,"model":78,"canonical":78,"role":79,"dataset":80,"specs":81,"locator":82},"tls_scanner","Riegl VZ-400","dataset sensor","lecturehall, campus, wrzburg","used by the authors to record the lecturehall, campus (146 scans, about 15 million points per scan) and wrzburg (6 scans, 86 million points) datasets","Sec. IX, Table I",[],{"totalRows":85,"groupCount":86,"groups":87,"others":480},30,4,[88,326,424,448],{"slug":89,"group":90,"sourceId":91,"sourceLabel":92,"table":93,"selfRows":94,"metrics":95,"seqs":105,"entrants":118,"cells":133,"outcomes":320,"locators":321,"hardware":322,"wordings":323,"notes":324},"erasor2021-table-ii","erasor2021:Table II","erasor2021","Lim et al., 2021","Table II",15,[96,100,102],{"label":97,"unit":98,"statistic":99,"alignment":48},"Preservation Rate (PR)","%","not_reported",{"label":101,"unit":98,"statistic":99,"alignment":48},"Rejection Rate (RR)",{"label":103,"unit":104,"statistic":99,"alignment":48},"F1 score","ratio",[106,110,112,114,116],{"dataset":107,"sequence":108,"environment":109},"SemanticKITTI","00 (frames 4390-4530)","urban driving (countryside, highway, intersections)",{"dataset":107,"sequence":111,"environment":109},"01 (frames 150-250)",{"dataset":107,"sequence":113,"environment":109},"02 (frames 860-950)",{"dataset":107,"sequence":115,"environment":109},"05 (frames 2350-2670)",{"dataset":107,"sequence":117,"environment":109},"07 (frames 630-820)",[119,123,125,126,129,131],{"name":120,"methodId":121,"linkable":122,"proposed":71,"self":71},"OctoMap - 0.05","hornung2013octomap",true,{"name":124,"methodId":121,"linkable":122,"proposed":71,"self":71},"OctoMap - 0.2",{"name":7,"methodId":5,"linkable":122,"proposed":71,"self":122},{"name":127,"methodId":128,"linkable":122,"proposed":71,"self":71},"Removert - RM3","removert2020",{"name":130,"methodId":128,"linkable":122,"proposed":71,"self":71},"Removert - RM3+RV1",{"name":132,"methodId":91,"linkable":122,"proposed":122,"self":71},"ERASOR (Ours)",[134,138,141,144,146,148,150,152,154,156,159,161,163,165,167,169,172,174,176,178,180,182,184,186,188,190,192,194,196,198,200,202,204,206,208,210,212,214,216,218,220,222,224,226,228,230,232,234,236,238,240,242,244,246,248,250,252,254,256,258,260,262,264,266,268,270,272,274,276,278,280,282,284,286,288,290,292,294,296,298,300,302,304,306,308,310,312,314,316,318],[135,135,135,136,137,135,137,137,135],0,76.731,-1,[135,139,135,140,137,135,137,137,135],1,99.124,[135,142,135,143,137,135,137,137,135],2,0.865,[139,135,135,145,137,135,137,137,135],34.568,[139,139,135,147,137,135,137,137,135],99.979,[139,142,135,149,137,135,137,137,135],0.514,[142,135,135,151,137,135,137,137,135],37.523,[142,139,135,153,137,135,137,137,135],89.116,[142,142,135,155,137,135,137,137,135],0.528,[157,135,135,158,137,135,137,137,135],3,85.502,[157,139,135,160,137,135,137,137,135],99.354,[157,142,135,162,137,135,137,137,135],0.919,[86,135,135,164,137,135,137,137,135],86.829,[86,139,135,166,137,135,137,137,135],90.617,[86,142,135,168,137,135,137,137,135],0.887,[170,135,135,171,137,135,137,137,135],5,93.98,[170,139,135,173,137,135,137,137,135],97.081,[170,142,135,175,137,135,137,137,135],0.955,[135,135,139,177,137,135,137,137,135],53.163,[135,139,139,179,137,135,137,137,135],99.663,[135,142,139,181,137,135,137,137,135],0.693,[139,135,139,183,137,135,137,137,135],20.777,[139,139,139,185,137,135,137,137,135],99.863,[139,142,139,187,137,135,137,137,135],0.344,[142,135,139,189,137,135,137,137,135],36.349,[142,139,139,191,137,135,137,137,135],93.116,[142,142,139,193,137,135,137,137,135],0.523,[157,135,139,195,137,135,137,137,135],94.221,[157,139,139,197,137,135,137,137,135],93.608,[157,142,139,199,137,135,137,137,135],0.939,[86,135,139,201,137,135,137,137,135],95.815,[86,139,139,203,137,135,137,137,135],57.077,[86,142,139,205,137,135,137,137,135],0.715,[170,135,139,207,137,135,137,137,135],91.487,[170,139,139,209,137,135,137,137,135],95.383,[170,142,139,211,137,135,137,137,135],0.934,[135,135,142,213,137,135,137,137,135],54.112,[135,139,142,215,137,135,137,137,135],98.769,[135,142,142,217,137,135,137,137,135],0.699,[139,135,142,219,137,135,137,137,135],23.746,[139,139,142,221,137,135,137,137,135],99.792,[139,142,142,223,137,135,137,137,135],0.384,[142,135,142,225,137,135,137,137,135],29.037,[142,139,142,227,137,135,137,137,135],94.527,[142,142,142,229,137,135,137,137,135],0.444,[157,135,142,231,137,135,137,137,135],76.319,[157,139,142,233,137,135,137,137,135],96.799,[157,142,142,235,137,135,137,137,135],0.853,[86,135,142,237,137,135,137,137,135],83.293,[86,139,142,239,137,135,137,137,135],88.371,[86,142,142,241,137,135,137,137,135],0.858,[170,135,142,243,137,135,137,137,135],87.731,[170,139,142,245,137,135,137,137,135],97.008,[170,142,142,247,137,135,137,137,135],0.921,[135,135,157,249,137,135,137,137,135],76.341,[135,139,157,251,137,135,137,137,135],96.785,[135,142,157,253,137,135,137,137,135],0.854,[139,135,157,255,137,135,137,137,135],33.904,[139,139,157,257,137,135,137,137,135],99.882,[139,142,157,259,137,135,137,137,135],0.506,[142,135,157,261,137,135,137,137,135],38.495,[142,139,157,263,137,135,137,137,135],90.631,[142,142,157,265,137,135,137,137,135],0.54,[157,135,157,267,137,135,137,137,135],86.9,[157,139,157,269,137,135,137,137,135],87.88,[157,142,157,271,137,135,137,137,135],0.874,[86,135,157,273,137,135,137,137,135],88.17,[86,139,157,275,137,135,137,137,135],79.981,[86,142,157,277,137,135,137,137,135],0.839,[170,135,157,279,137,135,137,137,135],88.73,[170,139,157,281,137,135,137,137,135],98.262,[170,142,157,283,137,135,137,137,135],0.933,[135,135,86,285,137,135,137,137,135],77.838,[135,139,86,287,137,135,137,137,135],96.938,[135,142,86,289,137,135,137,137,135],0.863,[139,135,86,291,137,135,137,137,135],38.183,[139,139,86,293,137,135,137,137,135],99.565,[139,142,86,295,137,135,137,137,135],0.552,[142,135,86,297,137,135,137,137,135],34.772,[142,139,86,299,137,135,137,137,135],91.983,[142,142,86,301,137,135,137,137,135],0.505,[157,135,86,303,137,135,137,137,135],80.689,[157,139,86,305,137,135,137,137,135],98.822,[157,142,86,307,137,135,137,137,135],0.888,[86,135,86,309,137,135,137,137,135],82.038,[86,139,86,311,137,135,137,137,135],95.504,[86,142,86,313,137,135,137,137,135],0.883,[170,135,86,315,137,135,137,137,135],90.624,[170,139,86,317,137,135,137,137,135],99.271,[170,142,86,319,137,135,137,137,135],0.948,[],[93],[],[],[325],"Static-map benchmark on five manually selected SemanticKITTI frame ranges with SuMa poses; PR and RR computed voxel-wise with 0.2 voxel size for all methods; OctoMap run at 0.05 and 0.2 voxel sizes; Removert RM3 = three removal stages, RM3+RV1 adds one revert stage",{"slug":327,"group":328,"sourceId":5,"sourceLabel":6,"table":329,"selfRows":330,"metrics":331,"seqs":340,"entrants":365,"cells":370,"outcomes":411,"locators":413,"hardware":414,"wordings":416,"notes":417},"schauer2018peopleremover-table-i","schauer2018peopleremover:Table I","Table I",12,[332,335,338],{"label":333,"unit":334,"statistic":99,"alignment":48},"F1-score (dynamic point classification)","fraction",{"label":336,"unit":337,"statistic":99,"alignment":48},"runtime (full pipeline)","s",{"label":336,"unit":339,"statistic":99,"alignment":48},"hours",[341,345,349,353,357,361],{"dataset":342,"sequence":343,"environment":344},"sim (Underwood et al.)","8 scans","synthetic scene",{"dataset":346,"sequence":347,"environment":348},"lab (Underwood et al.)","12 scans","lab scene (Underwood et al.), very noisy, small moving boxes",{"dataset":350,"sequence":351,"environment":352},"carpark (Underwood et al.)","4 scans","car park scene (Underwood et al.)",{"dataset":354,"sequence":355,"environment":356},"lecturehall (own, Riegl VZ-400)","2 scans","indoor lecture hall",{"dataset":358,"sequence":359,"environment":360},"campus (own, Riegl VZ-400)","146 scans","campus, large open spaces",{"dataset":362,"sequence":363,"environment":364},"wrzburg (own, Riegl VZ-400)","6 scans","outdoor marketplace",[366,368],{"name":367,"methodId":53,"linkable":71,"proposed":71,"self":71},"underwood",{"name":369,"methodId":5,"linkable":122,"proposed":122,"self":122},"3dtk (peopleremover)",[371,373,375,376,378,380,382,384,386,388,390,392,394,396,398,399,401,402,403,405,406,408,409],[135,135,135,372,137,135,137,137,135],0.98,[135,139,135,374,137,135,135,137,135],25,[139,135,135,372,137,135,137,137,135],[139,139,135,377,137,135,135,137,135],6,[135,135,139,379,137,135,137,137,139],0.71,[135,139,139,381,137,135,135,137,139],405,[139,135,139,383,137,135,137,137,139],0.42,[139,139,139,385,137,135,135,137,139],29,[135,135,142,387,137,135,137,137,142],0.78,[135,139,142,389,137,135,135,137,142],34,[139,135,142,391,137,135,137,137,142],0.83,[139,139,142,393,137,135,135,137,142],23,[135,135,157,395,137,135,137,137,157],0.96,[135,139,157,397,137,135,135,137,157],837,[139,135,157,395,137,135,137,137,157],[139,139,157,400,137,135,135,137,157],687,[135,135,86,53,135,135,137,137,86],[139,135,86,53,135,135,137,137,86],[139,142,86,404,137,135,135,137,86],13.1,[135,135,170,53,135,135,137,137,170],[135,139,170,407,137,135,135,137,170],7961,[139,135,170,53,135,135,137,137,170],[139,139,170,410,137,135,135,137,170],4967,[412],"not_applicable (no labels)",[329],[415],"not reported (single-threaded)",[],[418,419,420,421,422,423],"Best-parameter F1 and full-pipeline single-threaded runtime; Underwood et al. run on all scan pairs; no clustering or sub-voxel step; sim: 387,838 points, 8 scans, 28 pairs, Ta 1.4, Tr 0.1 m, voxel 0.6 m","Best-parameter F1 and full-pipeline single-threaded runtime; Underwood et al. run on all scan pairs; lab: 5,815,910 points, 12 scans, 66 pairs, Ta 1.2, Tr 0.2 m, voxel 0.175 m; 0.19% of points dynamic","Best-parameter F1 and full-pipeline single-threaded runtime; Underwood et al. run on all scan pairs; carpark: 1,965,017 points, 4 scans, 6 pairs, Ta 1.0, Tr 0.35 m, voxel 0.125 m","Best-parameter F1 and full-pipeline single-threaded runtime; lecturehall: 44,574,647 points, 2 scans, 1 pair, Ta 0.8, Tr 0.3 m, voxel 0.1 m; the fairest runtime comparison","Unlabelled campus dataset: 2,227,455,077 points, 146 scans, 3456 of 10585 scan pairs compared for Underwood et al. after an overlap heuristic; lecturehall parameters reused","Unlabelled wrzburg marketplace dataset: 86,585,411 points, 6 scans, 15 pairs, registered with slam6D; voxel 0.1 m",{"slug":425,"group":426,"sourceId":5,"sourceLabel":6,"table":427,"selfRows":142,"metrics":428,"seqs":431,"entrants":433,"cells":436,"outcomes":441,"locators":442,"hardware":444,"wordings":445,"notes":446},"schauer2018peopleremover-text-sec-ix","schauer2018peopleremover:Text Sec. IX","Text Sec. IX",[429,430],{"label":333,"unit":334,"statistic":99,"alignment":48},{"label":336,"unit":337,"statistic":99,"alignment":48},[432],{"dataset":354,"sequence":355,"environment":356},[434],{"name":435,"methodId":5,"linkable":122,"proposed":122,"self":122},"peopleremover, voxel 17.5 cm",[437,439],[135,135,135,438,137,135,137,137,135],0.95,[135,139,135,440,137,135,135,137,135],567,[],[443],"Sec. IX",[415],[],[447],"Quality-runtime trade-off on lecturehall with a 17.5 cm voxel instead of 10 cm",{"slug":449,"group":450,"sourceId":91,"sourceLabel":92,"table":451,"selfRows":139,"metrics":452,"seqs":455,"entrants":458,"cells":465,"outcomes":473,"locators":475,"hardware":476,"wordings":477,"notes":478},"erasor2021-table-iii","erasor2021:Table III","Table III",[453],{"label":454,"unit":337,"statistic":99,"alignment":48},"Runtime\u002Fiteration",[456],{"dataset":107,"sequence":111,"environment":457},"urban driving (highway)",[459,461,462,464],{"name":460,"methodId":121,"linkable":122,"proposed":71,"self":71},"OctoMap",{"name":7,"methodId":5,"linkable":122,"proposed":71,"self":122},{"name":463,"methodId":128,"linkable":122,"proposed":71,"self":71},"Removert",{"name":132,"methodId":91,"linkable":122,"proposed":122,"self":71},[466,468,469,471],[135,135,135,467,137,135,137,137,135],1.077,[139,135,135,53,135,135,137,137,135],[142,135,135,470,137,135,137,137,135],0.8307,[157,135,135,472,137,135,137,137,135],0.0732,[474],"other: printed as '1,000' s; decimal separator ambiguous",[451],[],[],[479],"Runtime per iteration of each dynamic-removal method on SemanticKITTI sequence 01; hardware not reported",[],1790510663670]