[{"data":1,"prerenderedAt":285},["ShallowReactive",2],{"method-pomerleau2014_icpmapper":3},{"method":4,"reference":57,"equipment":80,"figures":98,"results":99},{"id":5,"label":6,"shortName":7,"title":7,"year":8,"era":9,"cluster":10,"scope":11,"keyIdeaZh":12,"keyIdeaEn":13,"fulltextStatus":14,"publicationStatus":15,"recommendation":16,"constructionRelevance":17,"validationEnvironment":18,"strengths":20,"limitations":25,"sensors":32,"platform":35,"estimator":37,"association":38,"timeModel":39,"deskew":40,"loopClosure":40,"globalOptimization":41,"mapRepresentation":42,"prior":43,"outputGeometry":44,"compute":45,"codeUrl":46,"codeLicense":47,"relatedVersions":48},"pomerleau2014_icpmapper","Pomerleau et al., 2014","Long-term 3D map maintenance in dynamic environments",2014,"classic","C01","odometry_with_local_mapping","本文提出以單一 3D 雷射進行長期定位與建圖的系統，重點是地圖隨時間的維護。新點雲先以 libpointmatcher 的 ICP 配準到全域稀疏點雲地圖，系統再依可見性假設逐點更新地圖點為動態的貝氏機率：若新讀值出現在原地圖點後方（1 度圓錐內），代表雷射穿過了該點位置，該點較可能已移動；更新時同時考慮兩射線夾角、入射角與距離雜訊。被判為動態的點再以雙向非剛性 ICP 估計逐點速度，不需物件模型或分群；P(Dyn) 也可作為 ICP 權重，以減少動態點造成的定位漂移。","Long-term 3D laser mapping that updates a per-point Bayesian dynamic probability from visibility in a sparse ICP map, estimates per-point velocities with dual non-rigid ICP, and can use P(Dyn) to down-weight dynamic points in registration.","full_text_reviewed","peer_reviewed_published","supplementary","未在工地內部驗證，但在蘇黎世 1.3 km、跨七個月的三次調查中，動態元素發生次數地圖標出兩處部分占用街道的施工區（Sec. V.C；Fig. 10）。逐點動態機率與可見性更新的做法，可用於重複掃描的工地中區分人員、機具等移動物體與新施作的靜態構件；但週期性出現的物體（如固定位置停放的機具）會使靜態的定義變得模糊（Sec. III）（推論）。",[19],"controlled_experiment",[21,22,23,24],"Overall parking-lot segmentation error decreased from 20% to around 5% over nine surveys in three days against a night-time ground-truth map (Sec. V.A; Fig. 6).","Weighting ICP by P(Dyn) solved localization drift caused by dynamic points accumulating in a 2 m layer above the ground (Sec. V.A).","Median estimated speeds of 1.7 and 4.1 m\u002Fs for a walking and jogging pedestrian, and 1.9, 4.6 and 8.1 m\u002Fs for a minibus driven at about 2, 5.5 and 11 m\u002Fs; velocities estimated up to 22 m from the sensor (Sec. V.B).","On a 1.3 km Zurich route surveyed three times over seven months, dynamic-element occurrence maps highlighted two construction sites, a felled tree and a busy intersection (Sec. V.C; Fig. 10).",[26,27,28,29,30,31],"Periodic dynamic objects such as trams and cars in fixed parking spaces make 'static' ambiguous; points that reach P(Dyn) >= 0.9 cannot return to static (Sec. III).","Free space is modelled only where the laser previously returned a reading, not throughout the volume (Sec. I).","Points near the ground, often lower parts of cars, were misclassified, so the static-point error rose over time (Sec. V.A).","Velocity vectors are noisy at small displacements; the data rate was reduced to 8 Hz to handle 1.5 to 10 m\u002Fs (Sec. V.D).","Trams were poorly detected because only their front or rear reveals motion (Sec. V.C).","Minibus median speeds were lower than the targets because acceleration phases were included (Sec. V.B).",[33,34],"3D LiDAR (Velodyne HDL-32E)","wheel odometry (prior alignment for registration)",[36],"UGV (ARTOR, based on the LandShark system)","ICP registration of each point cloud to the global map with libpointmatcher, wheel odometry as prior; P(Dyn) can weight points in ICP to discount dynamic points (Sec. II; Sec. V.A)","ICP nearest neighbours via libnabo k-d trees; for dynamic inference, map points are associated with each new reading inside a 1 deg cone in spherical coordinates (Sec. III)","discrete poses (one per scan)","not_reported","none reported","sparse global point cloud with surface normals, timestamps and per-point Bayesian probability of being dynamic; a new point is added only if its nearest map point is farther than 0.3 m; time history of per-point velocities kept (Sec. III-V)","prior map from earlier sessions (the first survey is the exploration phase)","static-scene point map (P(Dyn) \u003C 0.5), dynamic points with velocity vectors, and maps of dynamic-element occurrence, average speed and heading (Figs. 5, 10, 11)","single laptop with a four-core Intel Core i7 and 4 GB RAM: registration 6 to 11 Hz, map maintenance about 2 Hz with maps up to 600,000 points, velocity estimation 0.03 s on average (Sec. V.D)",null,"not_applicable (the paper links no repository; registration builds on the published libpointmatcher [16])",[49,53],{"relation":50,"title":51,"doi_or_url":52},"repository_copy","HAL hal-01143106 (submitted version)","https:\u002F\u002Fhal.science\u002Fhal-01143106",{"relation":54,"title":55,"doi_or_url":56},"attributed_software","ethz-asl\u002Fethzasl_icp_mapping; cited to this paper by interactive_slam, but the repository does not cite it","https:\u002F\u002Fgithub.com\u002Fethz-asl\u002Fethzasl_icp_mapping",{"id":5,"kind":58,"shortName":7,"title":7,"authors":59,"year":8,"venue":65,"venueType":66,"publisher":67,"volumeIssuePages":68,"doi":69,"arxivId":46,"url":70,"firstPublicDate":71,"publicationStatus":15,"metadataStatus":72,"fulltextStatus":14,"era":9,"classicReason":73,"codeUrl":46,"cluster":10,"topics":74,"mdpi":75,"verification":76,"label":6,"fulltextRoute":77,"versionRead":78,"addedByCensus":79},"method",[60,61,62,63,64],"François Pomerleau","Philipp Krüsi","Francis Colas","Paul Furgale","Roland Siegwart","2014 IEEE International Conference on Robotics and Automation (ICRA), Hong Kong","conference","IEEE","pp. 3712-3719","10.1109\u002Ficra.2014.6907397","https:\u002F\u002Fdoi.org\u002F10.1109\u002FICRA.2014.6907397","2014-05","metadata_verified","principle reused: per-point visibility-based dynamic probability with long-term maintenance of a 3D laser map and per-point velocity estimation, a precursor of later dynamic-point removal and lifelong LiDAR mapping; used as a baseline under the name ethzasl_icp_mapping by interactive_slam [interactiveslam2021].",[10],false,"corrected","NTU institutional (curl)","IEEE Xplore version of record PDF (ICRA 2014, pp. 3712-3719, 8 pp.)",true,[81,87,92],{"category":82,"model":83,"canonical":83,"role":84,"dataset":46,"specs":85,"locator":86},"lidar","Velodyne HDL-32E","method input","roughly 70,000 points per 360 deg scan at 11 Hz; maximum range 80 m","Sec. V",{"category":88,"model":89,"canonical":89,"role":84,"dataset":46,"specs":90,"locator":91},"platform","ARTOR","maximum speed 3.5 m\u002Fs, typically 1 m\u002Fs in crowded environments; large sensor suite","Sec. V; Fig. 1",{"category":93,"model":94,"canonical":94,"role":95,"dataset":46,"specs":96,"locator":97},"compute","four-core Intel Core i7","compute for runtime","single laptop, 4 GB RAM","Sec. V.D",[],{"totalRows":100,"groupCount":101,"groups":102,"others":284},15,4,[103,167,206,250],{"slug":104,"group":105,"sourceId":5,"sourceLabel":6,"table":106,"selfRows":107,"metrics":108,"seqs":118,"entrants":133,"cells":136,"outcomes":155,"locators":156,"hardware":159,"wordings":160,"notes":161},"pomerleau2014-icpmapper-text-sec-v-b","pomerleau2014_icpmapper:Text Sec. V.B","Text Sec. V.B",6,[109,114],{"label":110,"unit":111,"statistic":112,"alignment":113},"median estimated speed","m\u002Fs","median","none",{"label":115,"unit":116,"statistic":117,"alignment":113},"distance from the sensor up to which velocities can be estimated","m","max",[119,123,125,127,129,131],{"dataset":120,"sequence":121,"environment":122},"authors' ARTOR HDL-32E data (remote street)","pedestrian, Walking","outdoor street",{"dataset":120,"sequence":124,"environment":122},"pedestrian, Jogging",{"dataset":120,"sequence":126,"environment":122},"minibus, Slow",{"dataset":120,"sequence":128,"environment":122},"minibus, Medium",{"dataset":120,"sequence":130,"environment":122},"minibus, Fast",{"dataset":120,"sequence":132,"environment":122},"all runs",[134],{"name":135,"methodId":5,"linkable":79,"proposed":79,"self":79},"dual non-rigid ICP velocity estimation",[137,141,144,147,150,152],[138,138,138,139,140,138,140,140,138],0,1.7,-1,[138,138,142,143,140,138,140,140,138],1,4.1,[138,138,145,146,140,138,140,140,142],2,1.9,[138,138,148,149,140,138,140,140,145],3,4.6,[138,138,101,151,140,138,140,140,148],8.1,[138,142,153,154,140,142,140,140,101],5,22,[],[157,158],"Sec. V.B; Fig. 9","Sec. V.B",[],[],[162,163,164,165,166],"Controlled street without traffic, robot parked, one moving object at a time; target speed not stated; for the minibus the acceleration phase is included, which lowers the median","Controlled street without traffic, robot parked, one moving object at a time; target speed about 2 m\u002Fs; for the minibus the acceleration phase is included, which lowers the median","Controlled street without traffic, robot parked, one moving object at a time; target speed 5.5 m\u002Fs; for the minibus the acceleration phase is included, which lowers the median","Controlled street without traffic, robot parked, one moving object at a time; target speed 11 m\u002Fs; for the minibus the acceleration phase is included, which lowers the median","Maximum range at which velocities could be estimated in the controlled test",{"slug":168,"group":169,"sourceId":170,"sourceLabel":171,"table":172,"selfRows":101,"metrics":173,"seqs":183,"entrants":188,"cells":191,"outcomes":200,"locators":201,"hardware":202,"wordings":203,"notes":204},"interactiveslam2021-table-i","interactiveslam2021:Table I","interactiveslam2021","Koide et al., 2021a","Table I",[174,177,179,181],{"label":175,"unit":116,"statistic":176,"alignment":40},"ATE [m], mean +\u002F- std = 5.248 +\u002F- 5.313","mean",{"label":178,"unit":116,"statistic":176,"alignment":40},"RTE(5m) [m], mean +\u002F- std = 0.038 +\u002F- 0.094",{"label":180,"unit":116,"statistic":176,"alignment":40},"RTE(50m) [m], mean +\u002F- std = 0.433 +\u002F- 1.279",{"label":182,"unit":116,"statistic":176,"alignment":40},"RTE(500m) [m], mean +\u002F- std = 2.168 +\u002F- 2.113",[184],{"dataset":185,"sequence":186,"environment":187},"PASCO Mobile Measurement System outdoor dataset (released via SMRT-AIST)","outdoor sequence (about 20 min)","outdoor, about 200 m x 400 m",[189],{"name":190,"methodId":5,"linkable":79,"proposed":75,"self":79},"ethzasl_icp_mapping [33]",[192,194,196,198],[138,138,138,193,140,138,140,140,138],5.248,[138,142,138,195,140,138,140,140,138],0.038,[138,145,138,197,140,138,140,140,138],0.433,[138,148,138,199,140,138,140,140,138],2.168,[],[172],[],[],[205],"PASCO Mobile Measurement System outdoor dataset (about 200 m x 400 m, 20 min); ground truth is the 3D LiDAR position tracked by static total stations (translation only); errors given as mean +\u002F- std; RTE uses a modified routine that aligns each sub-trajectory within the evaluation window; the proposed rows start from LOAM odometry with 5 m keyframes",{"slug":207,"group":208,"sourceId":5,"sourceLabel":6,"table":209,"selfRows":148,"metrics":210,"seqs":219,"entrants":228,"cells":235,"outcomes":240,"locators":242,"hardware":243,"wordings":245,"notes":246},"pomerleau2014-icpmapper-text-sec-v-d","pomerleau2014_icpmapper:Text Sec. V.D","Text Sec. V.D",[211,214,216],{"label":212,"unit":213,"statistic":40,"alignment":113},"registration module rate","Hz",{"label":215,"unit":213,"statistic":176,"alignment":113},"map maintenance rate (in average)",{"label":217,"unit":218,"statistic":176,"alignment":113},"velocity estimation time (in average)","s",[220,224,226],{"dataset":221,"sequence":222,"environment":223},"authors' ARTOR data","all experiments","urban",{"dataset":221,"sequence":225,"environment":223},"1.3 km survey",{"dataset":221,"sequence":227,"environment":122},"controlled street",[229,231,233],{"name":230,"methodId":5,"linkable":79,"proposed":79,"self":79},"registration module (libpointmatcher ICP)",{"name":232,"methodId":5,"linkable":79,"proposed":79,"self":79},"global map maintenance module",{"name":234,"methodId":5,"linkable":79,"proposed":79,"self":79},"velocity estimation module",[236,237,238],[138,138,138,46,138,138,138,140,138],[142,142,142,145,140,138,138,140,142],[145,145,145,239,140,138,138,140,145],0.03,[241],"range 6 to 11 Hz",[97],[244],"single laptop, four-core Intel Core i7, 4 GB RAM",[],[247,248,249],"Module rates with input run at recorded rate; registration downsamples points and uses wheel odometry as prior","Map maintenance including dynamic-element identification, maps up to 600,000 points on the 1.3 km survey","Velocity estimation time in the single dynamic object experiment",{"slug":251,"group":252,"sourceId":5,"sourceLabel":6,"table":253,"selfRows":145,"metrics":254,"seqs":260,"entrants":267,"cells":270,"outcomes":274,"locators":277,"hardware":279,"wordings":280,"notes":281},"pomerleau2014-icpmapper-text-sec-v-a","pomerleau2014_icpmapper:Text Sec. V.A","Text Sec. V.A",[255,258],{"label":256,"unit":257,"statistic":40,"alignment":113},"overall classification error (start of the series)","%",{"label":259,"unit":257,"statistic":40,"alignment":113},"overall classification error (end of the series)",[261,265],{"dataset":262,"sequence":263,"environment":264},"authors' ARTOR HDL-32E data (parking lot)","first survey","outdoor parking lot with bike path",{"dataset":262,"sequence":266,"environment":264},"last survey",[268],{"name":269,"methodId":5,"linkable":79,"proposed":79,"self":79},"Bayesian dynamic-point classification",[271,273],[138,138,138,272,138,138,140,140,138],20,[138,142,142,153,142,138,140,140,142],[275,276],"approximate (text: from 20%)","approximate (text: to around 5%)",[278],"Sec. V.A; Fig. 6",[],[],[282,283],"Hospital visitor parking lot, nine surveys over three days; classification compared with a night-time ground-truth map (static if a ground-truth point lies within 0.15 m)","Same series, after the last survey",[],1790510661712]