[{"data":1,"prerenderedAt":596},["ShallowReactive",2],{"method-yan2026tunnel":3},{"method":4,"reference":58,"equipment":80,"figures":120,"results":121},{"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":28,"sensors":35,"platform":39,"estimator":41,"association":42,"timeModel":43,"deskew":44,"loopClosure":45,"globalOptimization":46,"mapRepresentation":47,"prior":48,"outputGeometry":49,"compute":50,"codeUrl":51,"codeLicense":52,"relatedVersions":53},"yan2026tunnel","Yan et al., 2026a","Deep feature-enhanced LVIO (tunnel)","Deep feature-enhanced LiDAR-visual-inertial odometry for robust mapping in tunnel environment",2026,"recent","C07","odometry_with_local_mapping","此研究針對隧道幾何特徵稀疏、結構重複而導致光達里程計退化的問題，提出光達、視覺與慣性融合的里程計。光達端以曲率區分邊緣與平面特徵，採點對線與點對面配準；將資訊矩陣求逆得到共變異數後，分別對旋轉與平移子區塊做特徵分解，以經驗門檻判定六自由度中哪些方向退化。視覺端以 SuperPoint 搭配自適應門檻擷取特徵、以 LightGlue 匹配，並結合 IMU 預積分，但不做後端最佳化。條件式擴展卡爾曼濾波器只在偵測到退化時，以選擇矩陣保留退化方向上的視覺慣性資訊來更新狀態。系統只做里程計，不含迴圈閉合；驗證完全使用 MIT 校園隧道（KMCT）與 WHU-Helmet 隧道、地鐵兩組公開資料。","Detects 6-DoF degeneracy of feature-based LiDAR odometry in tunnels from the covariance of the rotation and translation blocks, and fuses a SuperPoint and LightGlue visual-inertial odometry only along the degenerate directions through a conditional EKF.","full_text_reviewed","peer_reviewed_published","main_body","論文以隧道營運維護與機器人巡檢為應用情境（摘要與引言明確提及營運維護），刊於本文目標期刊。驗證資料為 MIT 校園地下隧道（KMCT，八台 Clearpath Jackal 機器人，總長 6753 m）與 WHU-Helmet 的 Tunnel（790.32 m）與 Subway（854.24 m）序列，並非施工中隧道；作者未自行蒐集資料，軌跡真值由資料集提供，參考量測方式未在文中說明。",[20,21],"underground_or_tunnel","public_benchmark",[23,24,25,26,27],"Lowest average ATE on both datasets: 3.21 m on KMCT (Fast-LIO2 3.26 m) and 3.74 m on WHU-Helmet (BALM 4.46 m, a 16.14% ATE reduction that the abstract labels as RPE) (Tables 1 to 2, Sec. 4.3)","Average RPE 1.77% on KMCT and 1.93% on WHU-Helmet per 100 m, 7.33% and 17.87% lower than Fast-LIO2 (Sec. 4.3, Tables 1 to 2)","Lowest average Mean Map Entropy, -8.58 on KMCT and -7.03 on WHU-Helmet (Tables 3 to 4); 3σ voxel error 0.12 versus 0.18 for LVI-SAM on a KMCT partial map (Sec. 4.3, Fig. 9)","About 25 to 29 ms per frame, 13.98% faster on average than R3LIVE++ (Table 5)","Removing the degeneracy module raises ATE by 18.00% and RPE by 24.37% on three KMCT sequences (Sec. 5)",[29,30,31,32,33,34],"Dynamic objects in tunnels are not considered; multi-robot collaborative mapping is left to future work (Sec. 6)","Not best on every sequence: KMCT 07_ac2-005 ATE 2.89 m versus 2.70 m for Fast-LIO2, 07_sob-002 1.18 m versus 1.13 m for COIN-LIO, 07_tho-001 8.48 m versus 8.37 m for Fast-LIO2 (Table 1)","Each sequence was run five times and the best trial reported (Sec. 4.1)","Degeneracy thresholds are set empirically (Sec. 3.1.2)","Map quality is assessed mainly by MME, which captures local consistency only (Sec. 4.2.2)","(inference) Only public datasets are used, absolute ATE is several metres, and code is not released (data on request)",[36,37,38],"3D LiDAR (mechanical spinning model assumed in the LO derivation; Velodyne on KMCT, Livox on WHU-Helmet)","camera (Intel RealSense D455 RGB-D on KMCT; helmet cameras on WHU-Helmet)","IMU (preintegrated in the VIO)",[40],"no own platform; offline evaluation on public datasets recorded by Clearpath Jackal ground robots (KMCT) and a helmet-mounted rig (WHU-Helmet)","Conditional EKF: LiDAR odometry runs continuously; when covariance-based detection flags degenerate rotation or translation directions, a selection matrix keeps only the visual-inertial information along those directions and the state is updated with a FAST-LIO style Kalman gain; the VIO itself has no back-end optimization (Secs. 3.2, 3.3)","LiDAR: curvature from five horizontal neighbours on each side separates edge and planar points, matched point-to-line and point-to-plane to global feature maps and solved by Gauss-Newton; degeneracy from eigen-decomposition of the rotation and translation blocks of the inverted information matrix against empirical thresholds; visual: SuperPoint features with an adaptive score threshold and LightGlue matching inside an ORB-SLAM3 style tracking thread (Secs. 3.1 to 3.2)","discrete scan poses; constant-velocity camera prediction; IMU preintegration between frames with high-rate IMU propagation of the output pose (Secs. 3.1.1, 3.2)","linear interpolation of the inter-scan transform across the sweep by point index (Eq. 3, Sec. 3.1.1)","none; the authors argue loop closure is often infeasible in tunnels and choose an odometry design (Secs. 2.3, 3.3.1)","none (no factor graph or bundle adjustment)","point cloud map built by accumulating registered scans, with global edge and planar feature maps used for matching (Secs. 3, 3.1.1)","none (official pre-trained SuperPoint and LightGlue models; no prior map)","dense LiDAR point cloud map evaluated by Mean Map Entropy (local consistency only) and, on one KMCT sequence, by voxel error against the ground-truth map (Secs. 4.2.2, 4.3, Fig. 9)","Intel i9-12900H CPU, NVIDIA 3070 Ti GPU, 32 GB RAM, Ubuntu 20.04 with CUDA, cuDNN and ONNX inference: 25.4 to 29.0 ms per frame in total (VIO about 20 to 23 ms, LO about 5 to 6 ms, EKF about 0.1 ms) versus 28.8 to 35.8 ms for R3LIVE++ and 49.7 to 67.6 ms for LVI-SAM (Table 5)",null,"not_verified",[54],{"relation":55,"title":56,"doi_or_url":57},"preprint","Deep feature-enhanced LiDAR-visual-inertial odometry for robotic inspection and mapping in degraded tunnel environments (SSRN)","10.2139\u002Fssrn.5412661",{"id":5,"kind":59,"shortName":7,"title":8,"authors":60,"year":9,"venue":66,"venueType":67,"publisher":68,"volumeIssuePages":69,"doi":70,"arxivId":51,"url":71,"firstPublicDate":72,"publicationStatus":16,"metadataStatus":73,"fulltextStatus":15,"era":10,"classicReason":74,"codeUrl":51,"cluster":11,"topics":75,"mdpi":76,"verification":77,"label":6,"fulltextRoute":78,"versionRead":79,"addedByCensus":76},"method",[61,62,63,64,65],"Yi Yan","Limao Zhang","Qingqing Ye","Zhaoxiang Zhang","Minghui Sun","Automation in Construction","journal","Elsevier","184: 106825","10.1016\u002Fj.autcon.2026.106825","https:\u002F\u002Fapi.crossref.org\u002Fworks\u002F10.1016\u002Fj.autcon.2026.106825","2025 (SSRN preprint 10.2139\u002Fssrn.5412661, Crossref deposit 2025-08-28; exact posting date not verified)","metadata_verified","not_applicable",[11],false,"corrected","NTU institutional (Chrome)","version of record, Automation in Construction 184 (April 2026) article 106825, ScienceDirect HTML full text",[81,87,93,97,101,106,108,112,116],{"category":82,"model":83,"canonical":83,"role":84,"dataset":51,"specs":85,"locator":86},"compute","Intel i9-12900H CPU with NVIDIA 3070 Ti","compute for runtime","32 GB RAM, Ubuntu 20.04, CUDA 11.3, cuDNN 8.9.6, ONNX 1.16.3","Sec. 4.1",{"category":88,"model":89,"canonical":89,"role":90,"dataset":91,"specs":92,"locator":86},"platform","Clearpath Jackal mobile robots (eight)","dataset sensor","Kimera-Multi Campus-Tunnel (KMCT)","collectively travelled 6753 m in about 30 min in the MIT campus tunnel",{"category":94,"model":95,"canonical":95,"role":90,"dataset":91,"specs":96,"locator":86},"rgbd","Intel RealSense D455","RGB-D camera on each robot",{"category":98,"model":99,"canonical":99,"role":90,"dataset":91,"specs":100,"locator":86},"lidar","Velodyne LiDAR (model not_reported)","not_reported",{"category":88,"model":102,"canonical":102,"role":90,"dataset":103,"specs":104,"locator":105},"helmet (per the dataset name and the title of ref. [21])","WHU-Helmet (WHUH)","Wuhan University helmet-based multisensor dataset","Sec. 4.1, ref. [21]",{"category":98,"model":107,"canonical":107,"role":90,"dataset":103,"specs":100,"locator":86},"LIVOX LiDAR (model not_reported)",{"category":109,"model":110,"canonical":111,"role":90,"dataset":103,"specs":100,"locator":86},"imu","IMU (model not_reported)","IMU (model not reported)",{"category":113,"model":114,"canonical":115,"role":90,"dataset":103,"specs":100,"locator":86},"gnss","GNSS receiver (model not_reported)","GNSS (receiver model not reported)",{"category":117,"model":118,"canonical":118,"role":90,"dataset":103,"specs":119,"locator":86},"camera","cameras (models not_reported)","Tunnel sequence 12,304 images in 1403 s; Subway 15,685 images in 1580 s",[],{"totalRows":122,"groupCount":123,"groups":124,"others":575},35,8,[125,331,429,465],{"slug":126,"group":127,"sourceId":5,"sourceLabel":6,"table":128,"selfRows":129,"metrics":130,"seqs":140,"entrants":158,"cells":185,"outcomes":323,"locators":325,"hardware":326,"wordings":327,"notes":328},"yan2026tunnel-table-1","yan2026tunnel:Table 1","Table 1",9,[131,135,137],{"label":132,"unit":133,"statistic":134,"alignment":100},"ATE (m), RMSE of global absolute trajectory error","m","RMSE",{"label":136,"unit":133,"statistic":134,"alignment":100},"Average ATE (m)",{"label":138,"unit":139,"statistic":134,"alignment":100},"Average RPE (%) per 100 m","% per 100 m",[141,144,146,148,150,152,154,156],{"dataset":91,"sequence":142,"environment":143},"07_ac2-005","MIT campus subterranean tunnel, long and uniform (Kimera-Multi Campus-Tunnel; eight Clearpath Jackal robots, 6753 m in about 30 min)",{"dataset":91,"sequence":145,"environment":143},"07_acl-001",{"dataset":91,"sequence":147,"environment":143},"07_api-003",{"dataset":91,"sequence":149,"environment":143},"07_hat-002",{"dataset":91,"sequence":151,"environment":143},"07_sob-002",{"dataset":91,"sequence":153,"environment":143},"07_spl-007",{"dataset":91,"sequence":155,"environment":143},"07_tho-001",{"dataset":91,"sequence":157,"environment":143},"Average (7 sequences)",[159,163,166,169,172,175,178,181,183],{"name":160,"methodId":161,"linkable":162,"proposed":76,"self":76},"ORB-SLAM3","orbslam3_2021",true,{"name":164,"methodId":165,"linkable":162,"proposed":76,"self":76},"VINS-Mono","vinsmono2018",{"name":167,"methodId":168,"linkable":162,"proposed":76,"self":76},"LeGO-LOAM","legoloam2018",{"name":170,"methodId":171,"linkable":162,"proposed":76,"self":76},"LIO-SAM","liosam2020",{"name":173,"methodId":174,"linkable":162,"proposed":76,"self":76},"Fast-LIO2","fastlio2_2022",{"name":176,"methodId":177,"linkable":162,"proposed":76,"self":76},"LVI-SAM","lvisam2021",{"name":179,"methodId":180,"linkable":162,"proposed":76,"self":76},"COIN-LIO","coinlio2024",{"name":182,"methodId":51,"linkable":76,"proposed":76,"self":76},"VINS-FEN",{"name":184,"methodId":5,"linkable":162,"proposed":162,"self":162},"Proposed",[186,189,192,195,198,201,204,207,210,212,214,216,218,220,222,224,226,228,230,232,234,236,238,240,242,244,246,248,249,250,251,253,255,257,259,261,263,265,267,269,271,273,274,276,278,279,280,282,284,286,288,290,292,294,296,297,298,299,300,302,304,306,307,309,311,312,314,316,318,319,321],[187,187,187,51,187,187,188,188,187],0,-1,[190,187,187,191,188,187,188,188,187],1,3.96,[193,187,187,194,188,187,188,188,187],2,3.63,[196,187,187,197,188,187,188,188,187],3,3.41,[199,187,187,200,188,187,188,188,187],4,2.7,[202,187,187,203,188,187,188,188,187],5,2.77,[205,187,187,206,188,187,188,188,187],6,2.71,[208,187,187,209,188,187,188,188,187],7,3.05,[123,187,187,211,188,187,188,188,187],2.89,[187,187,190,213,188,187,188,188,187],3.99,[190,187,190,215,188,187,188,188,187],3.74,[193,187,190,217,188,187,188,188,187],3.38,[196,187,190,219,188,187,188,188,187],3.24,[199,187,190,221,188,187,188,188,187],2.62,[202,187,190,223,188,187,188,188,187],2.9,[205,187,190,225,188,187,188,188,187],2.66,[208,187,190,227,188,187,188,188,187],2.86,[123,187,190,229,188,187,188,188,187],2.41,[187,187,193,231,188,187,188,188,187],1.82,[190,187,193,233,188,187,188,188,187],1.49,[193,187,193,235,188,187,188,188,187],1.41,[196,187,193,237,188,187,188,188,187],1.33,[199,187,193,239,188,187,188,188,187],0.94,[202,187,193,241,188,187,188,188,187],1.18,[205,187,193,243,188,187,188,188,187],0.97,[208,187,193,245,188,187,188,188,187],1.2,[123,187,193,247,188,187,188,188,187],0.92,[187,187,196,51,187,187,188,188,187],[190,187,196,51,187,187,188,188,187],[193,187,196,51,187,187,188,188,187],[196,187,196,252,188,187,188,188,187],7.01,[199,187,196,254,188,187,188,188,187],5.3,[202,187,196,256,188,187,188,188,187],4.97,[205,187,196,258,188,187,188,188,187],5.06,[208,187,196,260,188,187,188,188,187],5.03,[123,187,196,262,188,187,188,188,187],4.95,[187,187,199,264,188,187,188,188,187],2.38,[190,187,199,266,188,187,188,188,187],1.93,[193,187,199,268,188,187,188,188,187],2.64,[196,187,199,270,188,187,188,188,187],1.46,[199,187,199,272,188,187,188,188,187],1.16,[202,187,199,266,188,187,188,188,187],[205,187,199,275,188,187,188,188,187],1.13,[208,187,199,277,188,187,188,188,187],1.28,[123,187,199,241,188,187,188,188,187],[187,187,202,51,187,187,188,188,187],[190,187,202,281,188,187,188,188,187],2.18,[193,187,202,283,188,187,188,188,187],2.58,[196,187,202,285,188,187,188,188,187],2.37,[199,187,202,287,188,187,188,188,187],1.72,[202,187,202,289,188,187,188,188,187],2.24,[205,187,202,291,188,187,188,188,187],1.91,[208,187,202,293,188,187,188,188,187],1.94,[123,187,202,295,188,187,188,188,187],1.63,[187,187,205,51,187,187,188,188,187],[190,187,205,51,187,187,188,188,187],[193,187,205,51,187,187,188,188,187],[196,187,205,51,187,187,188,188,187],[199,187,205,301,188,187,188,188,187],8.37,[202,187,205,303,188,187,188,188,187],9.09,[205,187,205,305,188,187,188,188,187],8.52,[208,187,205,51,187,187,188,188,187],[123,187,205,308,188,187,188,188,187],8.48,[199,190,208,310,188,187,188,188,190],3.26,[199,193,208,291,188,187,188,188,190],[202,190,208,313,188,187,188,188,190],3.58,[202,193,208,315,188,187,188,188,190],2.3,[205,190,208,317,188,187,188,188,190],3.28,[205,193,208,291,188,187,188,188,190],[123,190,208,320,188,187,188,188,190],3.21,[123,193,208,322,188,187,188,188,190],1.77,[324],"failed",[128],[],[],[329,330],"ATE (RMSE, m) on KMCT; each ROS bag run five times and the best trial reported; 'Failed' = failure to run","ATE (RMSE, m) on KMCT; each ROS bag run five times and the best trial reported; 'Failed' = failure to run; averages only for methods that ran on every sequence",{"slug":332,"group":333,"sourceId":5,"sourceLabel":6,"table":334,"selfRows":129,"metrics":335,"seqs":340,"entrants":360,"cells":367,"outcomes":422,"locators":423,"hardware":424,"wordings":426,"notes":427},"yan2026tunnel-table-5","yan2026tunnel:Table 5","Table 5",[336],{"label":337,"unit":338,"statistic":339,"alignment":74},"Time consumption (ms)","ms","mean",[341,344,346,348,350,352,354,356,358],{"dataset":91,"sequence":342,"environment":343},"ac2-005","",{"dataset":91,"sequence":345,"environment":343},"acl-001",{"dataset":91,"sequence":347,"environment":343},"api-003",{"dataset":91,"sequence":349,"environment":343},"hat-002",{"dataset":91,"sequence":351,"environment":343},"sob-002",{"dataset":91,"sequence":353,"environment":343},"spl-007",{"dataset":91,"sequence":355,"environment":343},"tho-001",{"dataset":103,"sequence":357,"environment":343},"Tunnel",{"dataset":103,"sequence":359,"environment":343},"Subway",[361,363,364],{"name":362,"methodId":5,"linkable":162,"proposed":162,"self":162},"This work (Sum of VIO, LO and EKF)",{"name":176,"methodId":177,"linkable":162,"proposed":76,"self":76},{"name":365,"methodId":366,"linkable":162,"proposed":76,"self":76},"R3LIVE++","r3livepp2024",[368,370,372,374,376,378,380,382,384,386,388,390,392,394,396,398,400,402,404,406,408,410,412,414,416,418,420],[187,187,187,369,188,187,187,188,187],27.51,[187,187,190,371,188,187,187,188,187],25.57,[187,187,193,373,188,187,187,188,187],25.39,[187,187,196,375,188,187,187,188,187],25.94,[187,187,199,377,188,187,187,188,187],25.99,[187,187,202,379,188,187,187,188,187],26.65,[187,187,205,381,188,187,187,188,187],25.6,[187,187,208,383,188,187,187,188,187],28.98,[187,187,123,385,188,187,187,188,187],28.66,[190,187,187,387,188,187,187,188,187],51.36,[190,187,190,389,188,187,187,188,187],59.77,[190,187,193,391,188,187,187,188,187],59.92,[190,187,196,393,188,187,187,188,187],51.09,[190,187,199,395,188,187,187,188,187],51.27,[190,187,202,397,188,187,187,188,187],53.3,[190,187,205,399,188,187,187,188,187],49.71,[190,187,208,401,188,187,187,188,187],65.22,[190,187,123,403,188,187,187,188,187],67.57,[193,187,187,405,188,187,187,188,187],30.53,[193,187,190,407,188,187,187,188,187],31.21,[193,187,193,409,188,187,187,188,187],30.14,[193,187,196,411,188,187,187,188,187],28.92,[193,187,199,413,188,187,187,188,187],29.05,[193,187,202,415,188,187,187,188,187],31.27,[193,187,205,417,188,187,187,188,187],28.79,[193,187,208,419,188,187,187,188,187],33.5,[193,187,123,421,188,187,187,188,187],35.83,[],[334],[425],"Intel i9-12900H CPU, NVIDIA 3070 Ti, 32 GB RAM, Ubuntu 20.04",[],[428],"Time consumption per frame; proposed method component times (VIO, LO, EKF) not extracted, only their sum",{"slug":430,"group":431,"sourceId":5,"sourceLabel":6,"table":432,"selfRows":205,"metrics":433,"seqs":437,"entrants":440,"cells":447,"outcomes":459,"locators":460,"hardware":461,"wordings":462,"notes":463},"yan2026tunnel-table-6","yan2026tunnel:Table 6","Table 6",[434,435],{"label":136,"unit":133,"statistic":134,"alignment":100},{"label":436,"unit":139,"statistic":134,"alignment":100},"Average RPE (%)",[438],{"dataset":91,"sequence":439,"environment":143},"Average (3 sequences)",[441,443,445],{"name":442,"methodId":5,"linkable":162,"proposed":76,"self":162},"Standard VIO + Base (VINS-Mono front end)",{"name":444,"methodId":5,"linkable":162,"proposed":76,"self":162},"Base + EKF (standard EKF without selection)",{"name":446,"methodId":5,"linkable":162,"proposed":162,"self":162},"Proposed method",[448,450,452,453,455,457],[187,187,187,449,188,187,188,188,187],1.84,[187,190,187,451,188,187,188,188,187],1.59,[190,187,187,451,188,187,188,188,187],[190,190,187,454,188,187,188,188,187],1.37,[193,187,187,456,188,187,188,188,187],1.24,[193,190,187,458,188,187,188,188,187],0.9,[],[432],[],[],[464],"Ablation of deep-feature VIO and CEKF; averages over KMCT 07_api-003, 07_sob-002 and 07_spl-007; per-sequence rows not extracted",{"slug":466,"group":467,"sourceId":5,"sourceLabel":6,"table":468,"selfRows":199,"metrics":469,"seqs":474,"entrants":482,"cells":501,"outcomes":568,"locators":569,"hardware":570,"wordings":571,"notes":572},"yan2026tunnel-table-2","yan2026tunnel:Table 2","Table 2",[470,472,473],{"label":471,"unit":133,"statistic":134,"alignment":100},"ATE (m)",{"label":136,"unit":133,"statistic":134,"alignment":100},{"label":138,"unit":139,"statistic":134,"alignment":100},[475,477,479],{"dataset":103,"sequence":357,"environment":476},"WHU-Helmet Tunnel sequence, 790.32 m, 1403 s",{"dataset":103,"sequence":359,"environment":478},"WHU-Helmet Subway sequence, 854.24 m, 1580 s",{"dataset":103,"sequence":480,"environment":481},"Average (Tunnel, Subway)","WHU-Helmet Tunnel and Subway",[483,484,485,488,491,494,495,497,498,499],{"name":160,"methodId":161,"linkable":162,"proposed":76,"self":76},{"name":164,"methodId":165,"linkable":162,"proposed":76,"self":76},{"name":486,"methodId":487,"linkable":162,"proposed":76,"self":76},"LOAM","loam2014",{"name":489,"methodId":490,"linkable":162,"proposed":76,"self":76},"BALM","balm2021",{"name":492,"methodId":493,"linkable":162,"proposed":76,"self":76},"LIO-Mapping","liomapping2019",{"name":173,"methodId":174,"linkable":162,"proposed":76,"self":76},{"name":496,"methodId":366,"linkable":162,"proposed":76,"self":76},"R3live++",{"name":179,"methodId":180,"linkable":162,"proposed":76,"self":76},{"name":182,"methodId":51,"linkable":76,"proposed":76,"self":76},{"name":500,"methodId":5,"linkable":162,"proposed":162,"self":162},"This work",[502,504,505,506,508,510,512,513,515,517,519,521,523,525,527,529,531,533,535,537,539,541,543,545,547,549,551,553,555,557,559,561,563,565,566,567],[187,187,187,503,188,187,188,188,187],5.92,[190,187,187,51,187,187,188,188,187],[193,187,187,51,187,187,188,188,187],[196,187,187,507,188,187,188,188,187],4.53,[199,187,187,509,188,187,188,188,187],5.95,[202,187,187,511,188,187,188,188,187],4.21,[205,187,187,254,188,187,188,188,187],[208,187,187,514,188,187,188,188,187],4.05,[123,187,187,516,188,187,188,188,187],4.39,[129,187,187,518,188,187,188,188,187],4.04,[187,187,190,520,188,187,188,188,187],6.25,[190,187,190,522,188,187,188,188,187],6.18,[193,187,190,524,188,187,188,188,187],10.37,[196,187,190,526,188,187,188,188,187],4.38,[199,187,190,528,188,187,188,188,187],7.92,[202,187,190,530,188,187,188,188,187],5.29,[205,187,190,532,188,187,188,188,187],8.51,[208,187,190,534,188,187,188,188,187],4.19,[123,187,190,536,188,187,188,188,187],4.77,[129,187,190,538,188,187,188,188,187],3.44,[187,190,193,540,188,187,188,188,190],6.09,[187,193,193,542,188,187,188,188,190],7.58,[196,190,193,544,188,187,188,188,190],4.46,[196,193,193,546,188,187,188,188,190],2.54,[199,190,193,548,188,187,188,188,190],6.94,[199,193,193,550,188,187,188,188,190],3.34,[202,190,193,552,188,187,188,188,190],4.75,[202,193,193,554,188,187,188,188,190],2.35,[205,190,193,556,188,187,188,188,190],6.91,[205,193,193,558,188,187,188,188,190],2.95,[208,190,193,560,188,187,188,188,190],4.12,[208,193,193,562,188,187,188,188,190],2.22,[123,190,193,564,188,187,188,188,190],4.58,[123,193,193,315,188,187,188,188,190],[129,190,193,215,188,187,188,188,190],[129,193,193,266,188,187,188,188,190],[324],[468],[],[],[573,574],"ATE (RMSE, m) on WHU-Helmet with dataset ground-truth trajectory; LiDAR baselines use variants adapted to the Livox configuration; 'Failed' = failure to run","ATE (RMSE, m) on WHU-Helmet with dataset ground-truth trajectory; LiDAR baselines use variants adapted to the Livox configuration; 'Failed' = failure to run; averages only for methods that ran on both sequences",[576,581,586,591],{"group":577,"slug":578,"sourceLabel":6,"table":579,"selfRows":199,"datasets":580},"yan2026tunnel:Table 7","yan2026tunnel-table-7","Table 7",[91],{"group":582,"slug":583,"sourceLabel":6,"table":584,"selfRows":190,"datasets":585},"yan2026tunnel:Table 3","yan2026tunnel-table-3","Table 3",[91],{"group":587,"slug":588,"sourceLabel":6,"table":589,"selfRows":190,"datasets":590},"yan2026tunnel:Table 4","yan2026tunnel-table-4","Table 4",[103],{"group":592,"slug":593,"sourceLabel":6,"table":594,"selfRows":190,"datasets":595},"yan2026tunnel:Text Sec. 4.3","yan2026tunnel-text-sec-4-3","Text Sec. 4.3",[91],1790510661176]