[{"data":1,"prerenderedAt":543},["ShallowReactive",2],{"method-saloam2021":3},{"method":4,"reference":53,"equipment":79,"figures":99,"results":141},{"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":26,"sensors":32,"platform":34,"estimator":36,"association":37,"timeModel":38,"deskew":39,"loopClosure":40,"globalOptimization":41,"mapRepresentation":42,"prior":43,"outputGeometry":44,"compute":45,"codeUrl":46,"codeLicense":47,"relatedVersions":48},"saloam2021","Li et al., 2021a","SA-LOAM","SA-LOAM: Semantic-aided LiDAR SLAM with Loop Closure",2021,"recent","C04","full_slam_with_global_correction","SA-LOAM 以開源的 F-LOAM 為基礎，先以預訓練的 RangeNet++ 為每個 LiDAR 點加上語意標籤，再把語意用在里程計與迴圈偵測兩處。里程計部分，邊緣與平面特徵只與相同語意的子地圖點配對，依類別分別降採樣以保留小物體，並要求地面平面法向量垂直、建物平面法向量水平，以剔除擬合不良的平面。迴圈部分，把每幀點雲聚成語意圖，以圖匹配網路評分候選，再以語意輔助 ICP 做幾何驗證，最後用 g2o 位姿圖最佳化，得到全域一致的語意地圖。","Semantic-aided LOAM (built on F-LOAM): RangeNet++ labels drive label-consistent edge and plane matching, class-wise downsampling and plane-orientation checks in the odometry, and a semantic-graph similarity network plus ICP verification closes loops in a g2o pose graph, yielding a globally consistent semantic map.","full_text_reviewed","peer_reviewed_published","background","論文只在 KITTI 與 Ford Campus 的車載道路資料上評估，沒有施工現場資料。它依賴在 SemanticKITTI 上訓練的語意分割與道路場景類別（地面、建物等），作者也指出換到 Ford 資料時分割品質明顯下降；因此在工地這種類別與外觀差異大的場景是否適用未經驗證。以地面垂直、建物水平的平面約束剔除錯誤平面的想法，可作為室內外結構化施工場景的參考，但屬推論。",[20,21],"public_benchmark","cross_site",[23,24,25],"Average KITTI relative translational error 0.76% for both Ours-ODOM and Ours-LOOP versus 1.27% for the F-LOAM baseline (Table II)","Loop closure lowers average KITTI ATE from 3.49 m to 2.34 m, the lowest average among ISC-LOAM, SuMa and SuMa++ (Table III)","On the unseen Ford Campus data it has the lowest ATE on both sequences and is the only method that closes the loops on sequence 02 (Table IV; Sec. IV-C; Fig. 6)",[27,28,29,30,31],"Semantic segmentation quality drops markedly on Ford data, which inevitably affects later stages (Sec. IV-C; Fig. 7)","Semantic weights are set equally; automatic weighting is left to future work (Sec. IV-A)","Loop closure mainly reduces rotational error and has less effect on translational RPE (Sec. IV-B)","Runtime is not reported although a GPU is used for segmentation and graph matching (Sec. IV-A) (inference)","Evaluated only on driving datasets (Sec. IV)",[33],"3D LiDAR only (Velodyne HDL-64E in KITTI and Ford Campus); per-point semantics from a pre-trained RangeNet++ network (Sec. IV-A)",[35],"vehicle (KITTI; Ford Campus)","F-LOAM based scan-to-submap optimization of point-to-line and point-to-plane distances with semantic-related weights (set equal in the experiments); loop constraints added to a g2o pose graph (Sec. III-B; Sec. III-C; Sec. IV-A)","LOAM-style edge and planar features matched only to submap points of the same semantic label (k-d tree, 5 neighbours), class-specific downsampling, and plane fits kept only when ground normals are vertical and building normals horizontal; submap from the last 20 frames (Sec. III-B; Table I)","discrete poses","not described","candidates within an odometry-drift-dependent distance (up to 64 random candidates), similarity scored by the authors' semantic-graph matching network (score above 0.95, top 5 kept), then geometric verification by semantic-assisted ICP to a submap (Sec. III-C; Table I)","pose graph optimization with g2o when a loop is detected (Sec. III-C)","edge and planar semantic feature submaps for odometry; global semantic point map plus a lightweight semantic graph map (at most 100 nodes of centre and label per frame) for loop search (Sec. III-A; Sec. III-C)","semantic segmentation network pre-trained on SemanticKITTI (RangeNet++), 12 of 19 classes kept; graph matching network from the authors' prior work (Sec. IV-A)","trajectory and globally consistent semantic point cloud map (Figs. 1 and 4)","Intel Core i7-9750H at 3.00 GHz, 16 GB RAM, NVIDIA GeForce GTX1080 8 GB; segmentation and graph matching run in PyTorch; runtime not reported (Sec. IV-A)",null,"not_applicable (no official code found)",[49],{"relation":50,"title":51,"doi_or_url":52},"preprint","SA-LOAM (arXiv v2, 'Accepted by ICRA-2021', CC BY 4.0)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2106.11516",{"id":5,"kind":54,"shortName":7,"title":8,"authors":55,"year":9,"venue":63,"venueType":64,"publisher":65,"volumeIssuePages":66,"doi":67,"arxivId":68,"url":69,"firstPublicDate":70,"publicationStatus":16,"metadataStatus":71,"fulltextStatus":15,"era":10,"classicReason":72,"codeUrl":46,"cluster":11,"topics":73,"mdpi":74,"verification":75,"label":6,"fulltextRoute":76,"versionRead":77,"addedByCensus":78},"method",[56,57,58,59,60,61,62],"Lin Li","Xin Kong","Xiangrui Zhao","Wanlong Li","Feng Wen","Hongbo Zhang","Yong Liu","2021 IEEE International Conference on Robotics and Automation (ICRA)","conference","IEEE","pp. 7627-7634","10.1109\u002Ficra48506.2021.9560884","2106.11516","https:\u002F\u002Fdoi.org\u002F10.1109\u002FICRA48506.2021.9560884","2021-05","metadata_verified","not_applicable",[11],false,"confirmed","arXiv","arXiv v2 (2021-07-01), 'Accepted by ICRA-2021'; IEEE version of record not read",true,[80,87,91,96],{"category":81,"model":82,"canonical":82,"role":83,"dataset":84,"specs":85,"locator":86},"lidar","Velodyne HDL-64E","dataset sensor","KITTI odometry","used in both KITTI and Ford Campus","Sec. IV-A",{"category":81,"model":82,"canonical":82,"role":83,"dataset":88,"specs":89,"locator":90},"Ford Campus Vision and Lidar Dataset","Ford sensor setting gives sparser projection images than KITTI","Sec. IV-A; Sec. IV-C",{"category":92,"model":93,"canonical":93,"role":94,"dataset":46,"specs":95,"locator":86},"compute","Intel Core i7-9750H","compute for runtime","3.00 GHz, 16 GB RAM",{"category":92,"model":97,"canonical":97,"role":94,"dataset":46,"specs":98,"locator":86},"NVIDIA GeForce GTX1080","8 GB GPU memory",[100,113,123,132],{"refId":5,"refLabel":6,"fig":101,"whatZh":102,"license":103,"licenseUrl":104,"sourceUrl":105,"src":106,"width":107,"height":108,"thumb":109,"thumbWidth":110,"thumbHeight":111,"modified":112},"Fig. 1","SA-LOAM 在 KITTI 序列 01 建立的語意地圖，不同顏色代表不同語意類別，放大點為語意圖節點，並示範一次迴圈閉合","CC BY 4.0","http:\u002F\u002Fcreativecommons.org\u002Flicenses\u002Fby\u002F4.0\u002F","https:\u002F\u002Farxiv.org\u002Fhtml\u002F2106.11516v2\u002Fdemo.png","\u002Ffigure-files\u002Fsaloam2021\u002Ffig-1.webp",966,597,"\u002Ffigure-files\u002Fsaloam2021\u002Ffig-1.thumb.webp",480,297,"converted to WebP",{"refId":5,"refLabel":6,"fig":114,"whatZh":115,"license":103,"licenseUrl":104,"sourceUrl":116,"src":117,"width":118,"height":119,"thumb":120,"thumbWidth":110,"thumbHeight":121,"modified":122},"Fig. 2","系統流程：原始點雲經語意分割後同時送入運動估計與迴圈偵測，偵測到迴圈時更新位姿圖並維護全域語意地圖","https:\u002F\u002Farxiv.org\u002Fhtml\u002F2106.11516v2\u002Fpipeline.png","\u002Ffigure-files\u002Fsaloam2021\u002Ffig-2.webp",1400,521,"\u002Ffigure-files\u002Fsaloam2021\u002Ffig-2.thumb.webp",179,"resized to at most 1400 px wide and converted to WebP",{"refId":5,"refLabel":6,"fig":124,"whatZh":125,"license":103,"licenseUrl":104,"sourceUrl":126,"src":127,"width":128,"height":129,"thumb":130,"thumbWidth":110,"thumbHeight":131,"modified":112},"Fig. 4","KITTI 測試序列 19 在迴圈閉合前後的軌跡與語意地圖","https:\u002F\u002Farxiv.org\u002Fhtml\u002F2106.11516v2\u002Fmap_19.png","\u002Ffigure-files\u002Fsaloam2021\u002Ffig-4.webp",1212,1588,"\u002Ffigure-files\u002Fsaloam2021\u002Ffig-4.thumb.webp",629,{"refId":5,"refLabel":6,"fig":133,"whatZh":134,"license":103,"licenseUrl":104,"sourceUrl":135,"src":136,"width":137,"height":138,"thumb":139,"thumbWidth":110,"thumbHeight":140,"modified":112},"Fig. 6","Ford 資料序列 02 各方法的軌跡，只有 Ours-LOOP 正確找到閉合迴圈","https:\u002F\u002Farxiv.org\u002Fhtml\u002F2106.11516v2\u002Fford_02.jpg","\u002Ffigure-files\u002Fsaloam2021\u002Ffig-6.webp",1007,801,"\u002Ffigure-files\u002Fsaloam2021\u002Ffig-6.thumb.webp",382,{"totalRows":142,"groupCount":143,"groups":144,"others":542},48,3,[145,378,481],{"slug":146,"group":147,"sourceId":5,"sourceLabel":6,"table":148,"selfRows":149,"metrics":150,"seqs":160,"entrants":186,"cells":206,"outcomes":372,"locators":373,"hardware":374,"wordings":375,"notes":376},"saloam2021-table-ii","saloam2021:Table II","Table II",26,[151,155,157],{"label":152,"unit":153,"statistic":154,"alignment":72},"relative translational error (%)","%","mean",{"label":156,"unit":153,"statistic":154,"alignment":72},"relative translational error (%), Average",{"label":158,"unit":159,"statistic":154,"alignment":72},"relative rotational error (deg\u002F100m), Average","deg\u002F100m",[161,164,166,168,170,172,174,176,178,180,182,184],{"dataset":84,"sequence":162,"environment":163},"00*","vehicle, road",{"dataset":84,"sequence":165,"environment":163},"01",{"dataset":84,"sequence":167,"environment":163},"02*",{"dataset":84,"sequence":169,"environment":163},"03",{"dataset":84,"sequence":171,"environment":163},"04",{"dataset":84,"sequence":173,"environment":163},"05*",{"dataset":84,"sequence":175,"environment":163},"06*",{"dataset":84,"sequence":177,"environment":163},"07*",{"dataset":84,"sequence":179,"environment":163},"08*",{"dataset":84,"sequence":181,"environment":163},"09*",{"dataset":84,"sequence":183,"environment":163},"10",{"dataset":84,"sequence":185,"environment":163},"Average 00-10",[187,190,193,196,199,202,204],{"name":188,"methodId":189,"linkable":78,"proposed":74,"self":74},"LOAM* (from [19])","loam2017_auro",{"name":191,"methodId":192,"linkable":78,"proposed":74,"self":74},"FLOAM","floam2021",{"name":194,"methodId":195,"linkable":78,"proposed":74,"self":74},"ISC-LOAM","iscloam2020",{"name":197,"methodId":198,"linkable":78,"proposed":74,"self":74},"SUMA","suma2018",{"name":200,"methodId":201,"linkable":78,"proposed":74,"self":74},"SUMA++","sumapp2019",{"name":203,"methodId":5,"linkable":78,"proposed":78,"self":78},"Ours-ODOM",{"name":205,"methodId":5,"linkable":78,"proposed":78,"self":78},"Ours-LOOP",[207,211,214,217,219,222,225,228,231,234,237,240,243,244,246,248,250,251,252,254,256,258,260,262,264,266,268,270,272,274,276,278,280,282,284,286,288,290,292,293,295,297,299,300,301,303,305,307,308,310,312,314,315,317,319,321,323,325,326,328,330,332,333,335,337,339,341,342,344,345,347,348,350,352,354,355,356,358,359,360,361,362,363,365,366,367,368,369,370,371],[208,208,208,209,210,208,210,210,208],0,0.78,-1,[208,208,212,213,210,208,210,210,208],1,1.43,[208,208,215,216,210,208,210,210,208],2,0.92,[208,208,143,218,210,208,210,210,208],0.86,[208,208,220,221,210,208,210,210,208],4,0.71,[208,208,223,224,210,208,210,210,208],5,0.57,[208,208,226,227,210,208,210,210,208],6,0.65,[208,208,229,230,210,208,210,210,208],7,0.63,[208,208,232,233,210,208,210,210,208],8,1.12,[208,208,235,236,210,208,210,210,208],9,0.77,[208,208,238,239,210,208,210,210,208],10,0.79,[208,212,241,242,210,208,210,210,208],11,0.84,[212,208,208,216,210,208,210,210,208],[212,208,212,245,210,208,210,210,208],2.8,[212,208,215,247,210,208,210,210,208],1.56,[212,208,143,249,210,208,210,210,208],1.09,[212,208,220,213,210,208,210,210,208],[212,208,223,239,210,208,210,210,208],[212,208,226,253,210,208,210,210,208],0.72,[212,208,229,255,210,208,210,210,208],0.54,[212,208,232,257,210,208,210,210,208],1.11,[212,208,235,259,210,208,210,210,208],1.28,[212,208,238,261,210,208,210,210,208],1.77,[212,212,241,263,210,208,210,210,208],1.27,[212,215,241,265,210,208,210,210,208],0.49,[215,208,208,267,210,208,210,210,208],1.02,[215,208,212,269,210,208,210,210,208],2.92,[215,208,215,271,210,208,210,210,208],1.67,[215,208,143,273,210,208,210,210,208],1.15,[215,208,220,275,210,208,210,210,208],1.5,[215,208,223,277,210,208,210,210,208],0.81,[215,208,226,279,210,208,210,210,208],0.76,[215,208,229,281,210,208,210,210,208],0.56,[215,208,232,283,210,208,210,210,208],1.2,[215,208,235,285,210,208,210,210,208],1.4,[215,208,238,287,210,208,210,210,208],1.87,[215,212,241,289,210,208,210,210,208],1.35,[215,215,241,291,210,208,210,210,208],0.52,[143,208,208,236,210,208,210,210,208],[143,208,212,294,210,208,210,210,208],11.15,[143,208,215,296,210,208,210,210,208],2.93,[143,208,143,298,210,208,210,210,208],1.25,[143,208,220,218,210,208,210,210,208],[143,208,223,281,210,208,210,210,208],[143,208,226,302,210,208,210,210,208],0.64,[143,208,229,304,210,208,210,210,208],0.47,[143,208,232,306,210,208,210,210,208],1.06,[143,208,235,239,210,208,210,210,208],[143,208,238,309,210,208,210,210,208],0.99,[143,212,241,311,210,208,210,210,208],1.95,[143,215,241,313,210,208,210,210,208],0.48,[220,208,208,227,210,208,210,210,208],[220,208,212,316,210,208,210,210,208],1.63,[220,208,215,318,210,208,210,210,208],3.54,[220,208,143,320,210,208,210,210,208],0.67,[220,208,220,322,210,208,210,210,208],0.34,[220,208,223,324,210,208,210,210,208],0.4,[220,208,226,304,210,208,210,210,208],[220,208,229,327,210,208,210,210,208],0.39,[220,208,232,329,210,208,210,210,208],1.01,[220,208,235,331,210,208,210,210,208],0.58,[220,208,238,320,210,208,210,210,208],[220,212,241,334,210,208,210,210,208],0.94,[220,215,241,336,210,208,210,210,208],0.28,[223,208,208,338,210,208,210,210,208],0.59,[223,208,212,340,210,208,210,210,208],1.89,[223,208,215,236,210,208,210,210,208],[223,208,143,343,210,208,210,210,208],0.87,[223,208,220,338,210,208,210,210,208],[223,208,223,346,210,208,210,210,208],0.45,[223,208,226,291,210,208,210,210,208],[223,208,229,349,210,208,210,210,208],0.41,[223,208,232,351,210,208,210,210,208],0.85,[223,208,235,353,210,208,210,210,208],0.68,[223,208,238,209,210,208,210,210,208],[223,212,241,279,210,208,210,210,208],[223,215,241,357,210,208,210,210,208],0.31,[226,208,208,338,210,208,210,210,208],[226,208,212,340,210,208,210,210,208],[226,208,215,239,210,208,210,210,208],[226,208,143,343,210,208,210,210,208],[226,208,220,338,210,208,210,210,208],[226,208,223,364,210,208,210,210,208],0.37,[226,208,226,291,210,208,210,210,208],[226,208,229,349,210,208,210,210,208],[226,208,232,242,210,208,210,210,208],[226,208,235,236,210,208,210,210,208],[226,208,238,209,210,208,210,210,208],[226,212,241,279,210,208,210,210,208],[226,215,241,336,210,208,210,210,208],[],[148],[],[],[377],"KITTI odometry 00-10; mean relative pose error over 100-800 m trajectories (rotation deg\u002F100m \u002F translation %); * marks sequences with loops; LOAM values quoted from its journal paper [19]; other baselines run with open-source code",{"slug":379,"group":380,"sourceId":5,"sourceLabel":6,"table":381,"selfRows":382,"metrics":383,"seqs":388,"entrants":398,"cells":404,"outcomes":475,"locators":476,"hardware":477,"wordings":478,"notes":479},"saloam2021-table-iii","saloam2021:Table III","Table III",16,[384],{"label":385,"unit":386,"statistic":387,"alignment":387},"Absolute Trajectory Error (m)","m","not_reported",[389,390,391,392,393,394,395,396],{"dataset":84,"sequence":162,"environment":163},{"dataset":84,"sequence":167,"environment":163},{"dataset":84,"sequence":173,"environment":163},{"dataset":84,"sequence":175,"environment":163},{"dataset":84,"sequence":177,"environment":163},{"dataset":84,"sequence":179,"environment":163},{"dataset":84,"sequence":181,"environment":163},{"dataset":84,"sequence":397,"environment":163},"Average",[399,400,401,402,403],{"name":194,"methodId":195,"linkable":78,"proposed":74,"self":74},{"name":197,"methodId":198,"linkable":78,"proposed":74,"self":74},{"name":200,"methodId":201,"linkable":78,"proposed":74,"self":74},{"name":203,"methodId":5,"linkable":78,"proposed":78,"self":78},{"name":205,"methodId":5,"linkable":78,"proposed":78,"self":78},[405,407,409,411,413,414,416,418,420,422,424,425,426,427,429,431,433,435,437,438,439,440,442,444,446,448,450,452,454,456,458,460,462,463,465,467,468,470,472,473],[208,208,208,406,210,208,210,210,208],1.6,[208,208,212,408,210,208,210,210,208],4.77,[208,208,215,410,210,208,210,210,208],2.49,[208,208,143,412,210,208,210,210,208],1.03,[208,208,220,281,210,208,210,210,208],[208,208,223,415,210,208,210,210,208],4.88,[208,208,226,417,210,208,210,210,208],2.31,[208,208,229,419,210,208,210,210,208],2.52,[212,208,208,421,210,208,210,210,208],1.14,[212,208,212,423,210,208,210,210,208],44.13,[212,208,215,218,210,208,210,210,208],[212,208,143,353,210,208,210,210,208],[212,208,220,324,210,208,210,210,208],[212,208,223,428,210,208,210,210,208],2.09,[212,208,226,430,210,208,210,210,208],3.88,[212,208,229,432,210,208,210,210,208],7.59,[215,208,208,434,210,208,210,210,208],1.17,[215,208,212,436,210,208,210,210,208],12.99,[215,208,215,302,210,208,210,210,208],[215,208,143,281,210,208,210,210,208],[215,208,220,364,210,208,210,210,208],[215,208,223,441,210,208,210,210,208],2.44,[215,208,226,443,210,208,210,210,208],1.19,[215,208,229,445,210,208,210,210,208],2.76,[143,208,208,447,210,208,210,210,208],5.14,[143,208,212,449,210,208,210,210,208],9.71,[143,208,215,451,210,208,210,210,208],3.04,[143,208,143,453,210,208,210,210,208],0.69,[143,208,220,455,210,208,210,210,208],0.53,[143,208,223,457,210,208,210,210,208],3.61,[143,208,226,459,210,208,210,210,208],1.74,[143,208,229,461,210,208,210,210,208],3.49,[220,208,208,309,210,208,210,210,208],[220,208,212,464,210,208,210,210,208],9.24,[220,208,215,466,210,208,210,210,208],0.75,[220,208,143,302,210,208,210,210,208],[220,208,220,469,210,208,210,210,208],0.36,[220,208,223,471,210,208,210,210,208],3.24,[220,208,226,283,210,208,210,210,208],[220,208,229,474,210,208,210,210,208],2.34,[],[381],[],[],[480],"KITTI sequences with loops; absolute trajectory error (m); statistic and alignment not stated",{"slug":482,"group":483,"sourceId":5,"sourceLabel":6,"table":484,"selfRows":226,"metrics":485,"seqs":487,"entrants":494,"cells":501,"outcomes":536,"locators":537,"hardware":538,"wordings":539,"notes":540},"saloam2021-table-iv","saloam2021:Table IV","Table IV",[486],{"label":385,"unit":386,"statistic":387,"alignment":387},[488,491,493],{"dataset":88,"sequence":489,"environment":490},"Seq01","vehicle, campus (unseen data)",{"dataset":88,"sequence":492,"environment":490},"Seq02",{"dataset":88,"sequence":397,"environment":490},[495,496,497,498,499,500],{"name":191,"methodId":192,"linkable":78,"proposed":74,"self":74},{"name":194,"methodId":195,"linkable":78,"proposed":74,"self":74},{"name":197,"methodId":198,"linkable":78,"proposed":74,"self":74},{"name":200,"methodId":201,"linkable":78,"proposed":74,"self":74},{"name":203,"methodId":5,"linkable":78,"proposed":78,"self":78},{"name":205,"methodId":5,"linkable":78,"proposed":78,"self":78},[502,504,506,508,510,512,514,516,518,520,522,524,526,527,529,531,532,534],[208,208,208,503,210,208,210,210,208],1.61,[208,208,212,505,210,208,210,210,208],7.63,[208,208,215,507,210,208,210,210,208],4.62,[212,208,208,509,210,208,210,210,208],2.3,[212,208,212,511,210,208,210,210,208],10.38,[212,208,215,513,210,208,210,210,208],6.34,[215,208,208,515,210,208,210,210,208],4.45,[215,208,212,517,210,208,210,210,208],75.65,[215,208,215,519,210,208,210,210,208],40.05,[143,208,208,521,210,208,210,210,208],4.22,[143,208,212,523,210,208,210,210,208],63.27,[143,208,215,525,210,208,210,210,208],33.75,[220,208,208,289,210,208,210,210,208],[220,208,212,528,210,208,210,210,208],9.02,[220,208,215,530,210,208,210,210,208],5.18,[223,208,208,289,210,208,210,210,208],[223,208,212,533,210,208,210,210,208],6.97,[223,208,215,535,210,208,210,210,208],4.16,[],[484],[],[],[541],"Ford Campus Vision and Lidar Dataset, models and parameters tuned on KITTI only; absolute trajectory error (m); statistic and alignment not stated",[],1790510663396]