[{"data":1,"prerenderedAt":484},["ShallowReactive",2],{"method-wildcat2022":3},{"method":4,"reference":62,"equipment":83,"figures":160,"results":161},{"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":25,"limitations":32,"sensors":38,"platform":41,"estimator":46,"association":47,"timeModel":48,"deskew":49,"loopClosure":50,"globalOptimization":51,"mapRepresentation":52,"prior":53,"outputGeometry":54,"compute":55,"codeUrl":56,"codeLicense":57,"relatedVersions":58},"wildcat2022","Ramezani et al., 2022","Wildcat","Wildcat: Online Continuous-Time 3D Lidar-Inertial SLAM",2022,"recent","C04","full_slam_with_global_correction","Wildcat 是 CSIRO 的線上 3D LiDAR 慣性 SLAM，其里程計是 Zebedee 等離線連續時間方法概念的即時實作：在固定長度的滑動時間視窗內，把點雲依位置與時間聚成多解析度橢球面元（surfel），以面元對面元的點到面型成本與 IMU 成本共同修正取樣位姿，再以三次 B-spline 內插得到高頻軌跡並重投影面元，藉此處理運動畸變。後端以六秒子地圖為節點做位姿圖最佳化，加入重力方向項並合併重疊節點，使計算量隨探索空間而非任務時間成長，亦支援多機去中心化建圖。","Wildcat makes Zebedee-style continuous-time surfel odometry online in a sliding window with IMU fusion and B-spline interpolation, and adds submap-based pose-graph optimization with node merging and multi-agent support.","full_text_reviewed","preprint","main_body","論文引言提及 LiDAR 慣性 SLAM 可作為營建等測量的低成本替代方案，但僅為動機陳述；實驗包含地下礦坑式賽道（以測量級雷射掃描地圖為參考）與含室內、隧道、樓梯的 QCAT 園區（以 60 個以上測量標靶為參考），未在營建工地驗證。",[20,21,22,23,24],"underground_or_tunnel","public_benchmark","independent_reference","cross_site","completed_building",[26,27,28,29,30,31],"DARPA SubT Final Event: point-wise comparison of the 40 cm-voxelised multi-agent map to the 1 cm surveyed map gave 3 cm mean and 5 cm std error, with more than 95% of points within 10 cm after fine alignment (Sec. VI-B, Fig. 8)","QCAT handheld sequences (~5 km each, 63 surveyed targets): mean absolute target error 0.42 m (FlatPack) and 0.34 m (SpinningPack), lower than LIO-SAM and FAST-LIO2 on the targets those systems reached before failing (Sec. VI-D, Table II)","Field tested on handheld, legged, tracked and car platforms (Sec. I, VII)","With loop closure enabled on MulRan DCC03, APE was slightly lower than LIO-SAM (Sec. VI-C; values shown only as box plots in Fig. 9)","One common voxel-filter parameter set for all datasets, whereas LIO-SAM and FAST-LIO2 were tuned per scenario for QCAT (Sec. VI-D)","Node merging reduced 1402 submaps to 213 pose-graph nodes (about 85 %) on QCAT SpinningPack, with PGO memory below 3 GiB (Sec. VI-E, Figs. 16, 18)",[33,34,35,36,37],"Onboard Jetson AGX Xavier odometry runs at only about 1-4 Hz (Sec. VI-E)","Future work needed on odometry resilience across more environments and on loop-closure detection (Sec. VII)","Technology is subject of a PCT patent application and no official code is released (Acknowledgement; no repository in paper)","On MulRan DCC03 Wildcat odometry drift (2.9 %, 0.01 deg\u002Fm) was slightly higher than LIO-SAM odometry (2.4 %, 0.009 deg\u002Fm) (Sec. VI-C)","QCAT accuracy relies on manually picked target centres and MSAC alignment, and the baselines are scored on smaller target subsets than Wildcat, so the Table II comparison is not on identical targets (Sec. VI-D, Table II)",[39,40],"3D LiDAR Velodyne VLP-16 in two configurations: servo-spun at 0.5 Hz on an inclined mount for 120 deg vertical FoV with the measurement rate set to 20 Hz (SpinningPack), or fixed with the native 30 deg vertical FoV (FlatPack; rate not stated); Ouster OS1-64 at 10 Hz with 120 m range in MulRan DCC03","IMU (9-DoF 3DM-CV5 at 100 Hz in the SpinningPack; FlatPack IMU model not stated)",[42,43,44,45],"handheld","legged (Boston Dynamics Spot)","tracked robot","vehicle","sliding-window continuous-time trajectory optimization (Gauss-Newton with Cauchy IRLS) over sample-pose corrections, followed by pose-graph optimization with a gravity-direction term (Cauchy IRLS) (Sec. IV-V)","multi-resolution surfels (voxel clustering by position and time, planarity-filtered ellipsoids); reciprocal kNN matching in a 7-D descriptor space; point-to-plane-type surfel cost weighted by lidar noise and surfel thickness (Sec. IV-A to IV-C)","continuous-time: cubic B-spline interpolation between corrected sample poses, linear interpolation in cost terms (Sec. IV-B, IV-C)","surfels re-projected with the continuously updated IMU-rate trajectory (Sec. IV-B2)","candidates by Mahalanobis-distance search or place recognition (e.g., Scan Context); point-to-plane ICP between surfel submaps, with global initialization when uncertainty is large; Mahalanobis gating (Sec. V-B2)","pose graph over submap nodes with node merging for overlapping submaps; supports decentralized multi-agent mapping (Sec. V)","local multi-resolution surfel maps bundled into six-second submaps (Sec. V-A)","none","globally optimized multi-agent point cloud map (e.g., submitted to DARPA SubT) and trajectory (Sec. VI-B)","odometry about 63.3 ms average (about 15 Hz) on an Intel Xeon W-10885M laptop CPU; about 1 to 4 Hz on the pack's NVIDIA Jetson AGX Xavier; odometry memory plateaus at about 500 MiB and PGO memory stays below 3 GiB on QCAT SpinningPack (Sec. VI-E, Figs. 15, 18)",null,"not_verified",[59],{"relation":16,"title":60,"doi_or_url":61},"arXiv 2205.12595 v1 (IEEE T-RO template header, no journal version found on 2026-09-25)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2205.12595",{"id":5,"kind":63,"shortName":7,"title":8,"authors":64,"year":9,"venue":73,"venueType":16,"publisher":73,"volumeIssuePages":74,"doi":56,"arxivId":75,"url":61,"firstPublicDate":76,"publicationStatus":16,"metadataStatus":77,"fulltextStatus":15,"era":10,"classicReason":78,"codeUrl":56,"cluster":11,"topics":79,"mdpi":80,"verification":81,"label":6,"fulltextRoute":73,"versionRead":82,"addedByCensus":80},"method",[65,66,67,68,69,70,71,72],"Milad Ramezani","Kasra Khosoussi","Gavin Catt","Peyman Moghadam","Jason Williams","Paulo Borges","Fred Pauling","Navinda Kottege","arXiv","arXiv:2205.12595","2205.12595","2022-05-25","metadata_verified","not_applicable",[11],false,"corrected","arXiv 2205.12595v1 (2022-05-25, the only version; IEEE T-RO template header; no peer-reviewed version found in Crossref on 2026-09-25)",[84,91,95,101,106,111,117,121,127,132,136,139,144,148,151,157],{"category":85,"model":86,"canonical":86,"role":87,"dataset":88,"specs":89,"locator":90},"lidar","Velodyne VLP-16","method input","DARPA SubT Final Event (Team CSIRO Data61); QCAT (SpinningPack)","measurement rate 20 Hz; mounted at an inclined angle on a servomotor spinning about the sensor z axis at 0.5 Hz, giving 120 deg vertical FoV (SpinningPack)","Sec. VI-A1, VI-A3; Table I",{"category":85,"model":86,"canonical":86,"role":87,"dataset":92,"specs":93,"locator":94},"QCAT (FlatPack)","fixed, vertical FoV 30 deg (FlatPack)","Sec. VI-A3; Table I",{"category":96,"model":97,"canonical":97,"role":87,"dataset":98,"specs":99,"locator":100},"imu","3DM-CV5","DARPA SubT Final Event; QCAT (SpinningPack)","9-DoF, angular velocity and linear acceleration at 100 Hz (SpinningPack); FlatPack IMU model not stated","Sec. VI-A1",{"category":102,"model":103,"canonical":103,"role":104,"dataset":98,"specs":105,"locator":100},"camera","RGB camera (four per SpinningPack, model not stated)","dataset sensor","part of the SpinningPack; not used in the described Wildcat pipeline",{"category":85,"model":107,"canonical":107,"role":104,"dataset":108,"specs":109,"locator":110},"Ouster OS1-64","MulRan (DCC03)","10 Hz, range 120 m, vehicle-mounted","Sec. VI-A2; Table I",{"category":112,"model":113,"canonical":113,"role":114,"dataset":108,"specs":115,"locator":116},"gnss","GPS (model not stated)","reference or ground truth","combined with fiber optic gyro and SLAM to give 6-DoF ground truth at 100 Hz","Sec. VI-A2",{"category":118,"model":119,"canonical":119,"role":114,"dataset":108,"specs":120,"locator":116},"other","fiber optic gyro (model not stated)","part of the MulRan ground-truth solution",{"category":122,"model":123,"canonical":123,"role":114,"dataset":124,"specs":125,"locator":126},"tls_scanner","survey-grade laser scanner (model not stated)","DARPA SubT Final Event","DARPA ground-truth point cloud used at 1 cm resolution; about 100 person-hours according to DARPA","Sec. VI-B",{"category":118,"model":128,"canonical":128,"role":114,"dataset":129,"specs":130,"locator":131},"surveyed targets (63; survey instrument not stated)","QCAT","scattered over indoor, outdoor, 3-storey office and mock-up tunnel areas","Sec. VI-A3, VI-D; Fig. 12",{"category":133,"model":134,"canonical":134,"role":87,"dataset":124,"specs":135,"locator":100},"platform","Boston Dynamics Spot (two robots)","legged robots carrying SpinningPacks",{"category":133,"model":137,"canonical":137,"role":87,"dataset":124,"specs":138,"locator":100},"BIA5 ATR tracked robot (two robots)","tracked robots carrying SpinningPacks",{"category":140,"model":141,"canonical":141,"role":87,"dataset":98,"specs":142,"locator":143},"mobile_scanner_device","SpinningPack (CSIRO perception pack)","spinning VLP-16, 3DM-CV5 IMU, four RGB cameras; robot-mounted at DARPA and hand-held at QCAT","Sec. VI-A1, VI-A3; Fig. 6",{"category":140,"model":145,"canonical":145,"role":87,"dataset":92,"specs":146,"locator":147},"FlatPack (CSIRO hand-held perception pack)","fixed VLP-16 with 30 deg vertical FoV","Sec. VI-A3; Fig. 6",{"category":133,"model":149,"canonical":149,"role":104,"dataset":108,"specs":150,"locator":116},"vehicle (MulRan data collection car, model not stated)","urban driving in Daejeon, sequence about 5 km",{"category":152,"model":153,"canonical":153,"role":154,"dataset":56,"specs":155,"locator":156},"compute","Intel Xeon W-10885M","compute for runtime","laptop CPU","Sec. VI-E",{"category":152,"model":158,"canonical":158,"role":154,"dataset":56,"specs":159,"locator":156},"NVIDIA Jetson AGX Xavier","perception pack onboard computer; odometry about 1 to 4 Hz",[],{"totalRows":162,"groupCount":163,"groups":164,"others":459},36,8,[165,260,353,413],{"slug":166,"group":167,"sourceId":5,"sourceLabel":6,"table":168,"selfRows":163,"metrics":169,"seqs":185,"entrants":192,"cells":202,"outcomes":253,"locators":255,"hardware":256,"wordings":257,"notes":258},"wildcat2022-table-ii","wildcat2022:Table II","Table II",[170,175,178,181],{"label":171,"unit":172,"statistic":173,"alignment":174},"absolute error mean","m","mean","control points",{"label":176,"unit":172,"statistic":177,"alignment":174},"absolute error RMSE","RMSE",{"label":179,"unit":172,"statistic":180,"alignment":174},"absolute error std","std",{"label":182,"unit":183,"statistic":184,"alignment":174},"# targets","count","not_reported",[186,190],{"dataset":187,"sequence":188,"environment":189},"QCAT (in-house)","FlatPack","QCAT campus, handheld, indoor office (3 storeys, stairs), outdoor open and forested areas, mock-up tunnel",{"dataset":187,"sequence":191,"environment":189},"SpinningPack",[193,197,200],{"name":194,"methodId":195,"linkable":196,"proposed":80,"self":80},"LIO-SAM [3]","liosam2020",true,{"name":198,"methodId":199,"linkable":196,"proposed":80,"self":80},"FAST-LIO2 [7]","fastlio2_2022",{"name":201,"methodId":5,"linkable":196,"proposed":196,"self":196},"Wildcat (ours)",[203,207,210,213,216,218,220,222,224,226,228,230,232,234,236,238,240,242,244,246,247,249,250,252],[204,204,204,205,204,204,206,206,204],0,0.92,-1,[204,208,204,209,204,204,206,206,204],1,1.33,[204,211,204,212,204,204,206,206,204],2,0.97,[204,214,204,215,204,204,206,206,204],3,41,[208,204,204,217,204,204,206,206,204],1.09,[208,208,204,219,204,204,206,206,204],1.38,[208,211,204,221,204,204,206,206,204],0.85,[208,214,204,223,204,204,206,206,204],53,[211,204,204,225,206,204,206,206,204],0.42,[211,208,204,227,206,204,206,206,204],0.46,[211,211,204,229,206,204,206,206,204],0.19,[211,214,204,231,206,204,206,206,204],63,[204,204,208,233,204,204,206,206,204],1.69,[204,208,208,235,204,204,206,206,204],2.52,[204,211,208,237,204,204,206,206,204],1.9,[204,214,208,239,204,204,206,206,204],38,[208,204,208,241,204,204,206,206,204],0.43,[208,208,208,243,204,204,206,206,204],0.64,[208,211,208,245,204,204,206,206,204],0.47,[208,214,208,223,204,204,206,206,204],[211,204,208,248,206,204,206,206,204],0.34,[211,208,208,227,206,204,206,206,204],[211,211,208,251,206,204,206,206,204],0.31,[211,214,208,231,206,204,206,206,204],[254],"partial: failed before completing the sequence; statistics cover only identified targets",[168],[],[],[259],"QCAT surveyed targets (63 in total): point-to-point error between manually picked target centres in each map and surveyed positions after MSAC robust alignment; RMSE computed during alignment; LIO-SAM and FAST-LIO2 only on the subset of targets identified before they failed (mainly in the tunnel); baselines had tuned voxel-filter parameters, Wildcat used one common setting",{"slug":261,"group":262,"sourceId":263,"sourceLabel":264,"table":265,"selfRows":163,"metrics":266,"seqs":270,"entrants":295,"cells":302,"outcomes":345,"locators":346,"hardware":348,"wordings":349,"notes":350},"zhang2023hiltioxford-fig-9-printed-table","zhang2023hiltioxford:Fig. 9 (printed table)","zhang2023hiltioxford","Zhang et al., 2023c","Fig. 9 (printed table)",[267],{"label":268,"unit":269,"statistic":177,"alignment":184},"RMSE ATE [cm]","cm",[271,275,278,280,283,286,289,292],{"dataset":272,"sequence":273,"environment":274},"Hilti-Oxford (Hilti SLAM Challenge 2022)","Exp11 Lower Gallery","Sheldonian Theatre",{"dataset":272,"sequence":276,"environment":277},"Exp01 Construction Ground Level","live construction site",{"dataset":272,"sequence":279,"environment":277},"Exp02 Construction Multilevel",{"dataset":272,"sequence":281,"environment":282},"Exp21 Outside Building","Sheldonian exterior and quad",{"dataset":272,"sequence":284,"environment":285},"Exp03 Construction Stairs","live construction site, stairs, dark basement",{"dataset":272,"sequence":287,"environment":288},"Exp07 Long Corridor","100 m office corridor",{"dataset":272,"sequence":290,"environment":291},"Exp15 Attic to Upper Gallery","Sheldonian Theatre, narrow spaces",{"dataset":272,"sequence":293,"environment":294},"Exp09 Cupola","Sheldonian Theatre, narrow staircases",[296,298,300],{"name":297,"methodId":5,"linkable":196,"proposed":80,"self":196},"CSIRO (Wildcat SLAM)",{"name":299,"methodId":56,"linkable":80,"proposed":80,"self":80},"Vision&Robotics (MC2SLAM)",{"name":301,"methodId":56,"linkable":80,"proposed":80,"self":80},"HKU (FastLIO2, BALM)",[303,305,306,307,309,311,314,317,319,321,323,324,325,327,329,331,333,334,335,336,337,339,341,343],[204,204,204,304,206,204,206,206,204],0.9,[204,204,208,208,206,204,206,206,208],[204,204,211,237,206,204,206,206,208],[204,204,214,308,206,204,206,206,208],1.5,[204,204,310,304,206,204,206,206,208],4,[204,204,312,313,206,204,206,206,208],5,3.7,[204,204,315,316,206,204,206,206,208],6,2.7,[204,204,318,310,206,204,206,206,208],7,[208,204,204,320,206,204,206,206,208],0.7,[208,204,208,322,206,204,206,206,208],1.1,[208,204,211,211,206,204,206,206,208],[208,204,214,322,206,204,206,206,208],[208,204,310,326,206,204,206,206,208],4.8,[208,204,312,328,206,204,206,206,208],5.1,[208,204,315,330,206,204,206,206,208],6.6,[208,204,318,332,206,204,206,206,208],10.1,[211,204,204,304,206,204,206,206,208],[211,204,208,208,206,204,206,206,208],[211,204,211,211,206,204,206,206,208],[211,204,214,328,206,204,206,206,208],[211,204,310,338,206,204,206,206,208],2.1,[211,204,312,340,206,204,206,206,208],4.6,[211,204,315,342,206,204,206,206,208],13.9,[211,204,318,344,206,204,206,206,208],17.8,[],[347],"Fig. 9",[],[],[351,352],"Per-sequence RMSE ATE (cm) of the top three teams, printed as numbers inside Fig. 9; bold marks sub-cm results","See first Fig. 9 row",{"slug":354,"group":355,"sourceId":5,"sourceLabel":6,"table":356,"selfRows":318,"metrics":357,"seqs":376,"entrants":381,"cells":386,"outcomes":399,"locators":403,"hardware":407,"wordings":410,"notes":411},"wildcat2022-text-sec-vi-e","wildcat2022:Text Sec.VI-E","Text Sec.VI-E",[358,361,364,366,369,372,374],{"label":359,"unit":360,"statistic":173,"alignment":184},"average runtime of the main odometry optimisation loop","ms",{"label":362,"unit":363,"statistic":173,"alignment":184},"odometry rate (approximately 15 Hz)","Hz",{"label":365,"unit":363,"statistic":184,"alignment":184},"odometry rate on the onboard computer",{"label":367,"unit":368,"statistic":184,"alignment":184},"odometry memory plateau","MiB",{"label":370,"unit":371,"statistic":184,"alignment":184},"total PGO memory for the entire dataset","GiB",{"label":373,"unit":183,"statistic":184,"alignment":184},"total number of generated submaps",{"label":375,"unit":183,"statistic":184,"alignment":184},"pose-graph nodes after merging (about 85 % reduction)",[377,379],{"dataset":187,"sequence":191,"environment":378},"campus indoor and outdoor",{"dataset":380,"sequence":380,"environment":380},"not stated",[382,384],{"name":383,"methodId":5,"linkable":196,"proposed":196,"self":196},"Wildcat odometry",{"name":385,"methodId":5,"linkable":196,"proposed":196,"self":196},"Wildcat PGO",[387,389,391,392,394,395,397],[204,204,204,388,206,204,204,206,204],63.3,[204,208,204,390,204,208,204,206,204],15,[204,211,208,56,208,208,208,206,204],[204,214,204,393,204,211,204,206,204],500,[208,310,204,214,211,211,204,206,204],[208,312,204,396,206,214,204,206,204],1402,[208,315,204,398,206,214,204,206,204],213,[400,401,402],"approximate","range: about 1 to 4 Hz, stated as fast enough to finish each window before the next","upper bound: reported as less than 3 GiB",[404,156,405,406],"Sec. VI-E, Fig. 15","Sec. VI-E, Fig. 18","Sec. VI-E, Fig. 16",[408,409],"laptop with Intel Xeon W-10885M CPU","NVIDIA Jetson AGX Xavier (perception pack onboard computer)",[],[412],"Runtime and memory of Wildcat running online on QCAT SpinningPack",{"slug":414,"group":415,"sourceId":5,"sourceLabel":6,"table":416,"selfRows":312,"metrics":417,"seqs":429,"entrants":433,"cells":438,"outcomes":448,"locators":450,"hardware":455,"wordings":456,"notes":457},"wildcat2022-text-sec-vi-b","wildcat2022:Text Sec.VI-B","Text Sec.VI-B",[418,420,422,425,427],{"label":419,"unit":172,"statistic":173,"alignment":184},"average distance error between corresponding points",{"label":421,"unit":172,"statistic":180,"alignment":184},"standard deviation of distance error",{"label":423,"unit":424,"statistic":184,"alignment":184},"share of corresponding points with distance below 10 cm","%",{"label":426,"unit":424,"statistic":184,"alignment":184},"DARPA deviation (share of submitted points farther than 1 m from the survey map)",{"label":428,"unit":424,"statistic":184,"alignment":184},"DARPA coverage of the course",[430],{"dataset":124,"sequence":431,"environment":432},"prize run (multi-agent)","Louisville Mega Cavern, subterranean course with dust and smoke",[434,436],{"name":435,"methodId":5,"linkable":196,"proposed":196,"self":196},"Wildcat multi-agent map",{"name":437,"methodId":5,"linkable":196,"proposed":196,"self":196},"Wildcat multi-agent map (DARPA scoring)",[439,441,443,445,446],[204,204,204,440,206,204,206,206,204],0.03,[204,208,204,442,206,204,206,206,204],0.05,[204,211,204,444,204,208,206,206,204],95,[208,214,204,204,206,211,206,206,204],[208,310,204,447,206,214,206,206,204],91,[449],"lower bound: reported as more than 95 %",[451,452,453,454],"Sec. VI-B, Fig. 7","Sec. VI-B, Fig. 8","Sec. I, Sec. VI-B, Fig. 7 caption","Sec. I, Fig. 7 caption",[],[],[458],"DARPA SubT Final Event prize run, four robots with SpinningPacks: Wildcat map voxelised at 40 cm compared point-wise with the 1 cm DARPA survey map after a fine alignment (method not specified); DARPA's own scoring quoted separately",[460,467,473,477],{"group":461,"slug":462,"sourceLabel":463,"table":464,"selfRows":214,"datasets":465},"ebadi2024subt:Text Sec. IV-C","ebadi2024subt-text-sec-iv-c","Ebadi et al., 2024","Text Sec. IV-C",[124,466],"DARPA SubT Urban Event",{"group":468,"slug":469,"sourceLabel":6,"table":470,"selfRows":211,"datasets":471},"wildcat2022:Text Sec.VI-C","wildcat2022-text-sec-vi-c","Text Sec.VI-C",[472],"MulRan",{"group":474,"slug":475,"sourceLabel":264,"table":168,"selfRows":211,"datasets":476},"zhang2023hiltioxford:Table II","zhang2023hiltioxford-table-ii",[272],{"group":478,"slug":479,"sourceLabel":480,"table":481,"selfRows":208,"datasets":482},"okvis2x2025:Table IV","okvis2x2025-table-iv","Boche et al., 2025","Table IV",[483],"Hilti-Oxford (Hilti 2022 challenge)",1790510660668]