[{"data":1,"prerenderedAt":232},["ShallowReactive",2],{"method-zlot_bosse2014_mine":3},{"method":4,"reference":63,"equipment":82,"figures":115,"results":116},{"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":21,"limitations":29,"sensors":40,"platform":44,"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},"zlot_bosse2014_mine","Zlot & Bosse, 2014","CSIRO underground mine CT-SLAM (Northparkes)","Efficient Large‐scale Three‐dimensional Mobile Mapping for Underground Mines",2014,"classic","C11b","full_slam_with_global_correction","本文把 CSIRO 的連續時間非剛性配準（原始版本出自 bosse_zlot2009_ctscan）擴展成完整的地下礦坑建圖流程：旋轉 SICK LMS 291 與 MEMS IMU 裝在皮卡車斗上，於澳洲 Northparkes 銅金礦以一般行車速度行駛 17.1 公里（含停車共 1 小時 53 分）。處理分三段：先以滑動視窗的非剛性配準（面元匹配加 IMU 約束，修正量以 B-spline 表示）產生開迴路軌跡；再以關鍵點投票的場所辨識找出回溯路段的迴圈，經穩健位姿圖最佳化粗對齊後，對整條軌跡做非剛性配準；最後把點雲配準到礦方既有的測量剖面，以定位到地理座標並修正殘餘漂移。作者報告總處理時間 53.9 分鐘，少於擷取時間一半；開迴路軌跡相對於「已配準到測量圖」的軌跡，前進方向偏差約 0.2%；配準後 95% 面元距測量面元 50 cm 內（RMS 25 cm）。須注意這些與測量圖的比較是在把點雲配準到同一份測量圖之後計算，不是獨立驗證。","Extends CSIRO continuous-time non-rigid registration into a full mine-mapping pipeline (sliding-window lidar-inertial registration, keypoint-voting place recognition, robust pose-graph optimization, whole-trajectory refinement and registration to a mine survey); 17.1 km mapped in 113 min at Northparkes and processed in 53.9 min, with about 0.2% forward drift bias and a 25 cm RMS survey residual that are computed against the survey-registered solution rather than an independent reference.","full_text_reviewed","peer_reviewed_published","background","驗證場域為澳洲 Northparkes 銅金礦營運中的斜坡道與平巷（Sec. 3.1），屬地下工程與隧道類情境，不是營建工地。作業目的具工程任務性：礦方需以三維表面模型評估大型設備運入的淨空（Sec. 1、4.5），作者稱模型已被用於規劃，但未量測任務成果。與測量圖的比較是在點雲已配準到同一份測量圖之後計算，測量圖的量測方式與日期不明，且礦坑已有變動（Sec. 4.3），所以 RMS 25 cm 等數值只能視為配準殘差，不能當作獨立幾何精度。作者也指出壁面平滑的道路隧道或煤礦可能缺乏沿隧道方向的約束（Sec. 5），這是隧道施工應用的重要限制。",[20],"underground_or_tunnel",[22,23,24,25,26,27,28],"17.1 km of decline and drive mapped in 1 h 53 min (including 45 min of stops) at 20-30 km\u002Fh without disrupting mine operations (Sec. 3.1, 6)","Whole pipeline 53.9 min against 113 min acquisition; open-loop stage about 31% of acquisition time, about 20% while moving (Table I, Sec. 4.1)","Place recognition found loop closures with no false positives on this data set; robust pose-graph optimization tolerated 46.2% artificially injected false positives and about 93% false positives in survey matching (Sec. 4.2-4.3, 5)","Open-loop drift against the survey-registered trajectory: near-zero rotational bias with interquartile widths of a few hundredths of a degree; translational bias about 0.2% forward and 0.1% vertical (Sec. 4.4, Fig. 10)","After registration to the mine survey, 95% of map surfels lie within 50 cm of the corresponding survey surfels, RMS 25 cm (Sec. 4.3, Fig. 11)","Tight IMU constraints reduced the variance of rotational differences by an order of magnitude compared with the earlier FSR processing (Sec. 5)","Surface model reported to be used by mine operators for planning a major equipment transport through the decline (Sec. 1, 6)",[30,31,32,33,34,35,36,37,38,39],"Survey residuals (95% within 50 cm, RMS 25 cm) and drift statistics are computed against the survey-registered solution, i.e., after the survey was used as fixed registration targets; the authors call this an evaluation 'to a limited degree' (Sec. 4.3-4.4)","Survey quality uncertain: apparatus, method and date unknown; the mine changed after the survey; survey height above floor unknown and walls assumed planar and vertical (Sec. 4.3)","No topological loops in the mine: drift along long branches cannot be reliably corrected, and small yaw errors produce large offsets at branch ends (Sec. 4.2, 4.4, 6)","Residual accelerometer biases, which were not corrected, are named as the most likely source of the about 0.2% forward and 0.1% vertical open-loop drift biases; laser range bias or scale errors (range bias is not modelled in the calibration), mount calibration imperfections and survey errors are listed only as possible contributors that the authors believe to be small (Sec. 3.2, 4.4)","The tilted spin axis produced asymmetric scan density that can bias trajectory estimates at speed (Sec. 5)","Feature-poor tunnels (road tunnels, coal mines) may lose along-tunnel observability; authors suggest independent velocity sensing (Sec. 5)","Two data gaps of about 25 s from a loose ethernet cable produced jumps in the open-loop trajectory that were repaired only in later stages (Sec. 4.1, Fig. 6)","Vertical-laser timestamps suffered dropped frames from the SICK Toolbox driver (CPU load, USB buffering), leaving residual timing irregularities in the surface reconstruction (Sec. 5)","Surface model has a 160 mm blind-spot gap between the two vertical lasers and long-triangle artefacts at occlusion boundaries; mist returns had to be filtered (Sec. 3.1, 4.5)","Single mine site; MATLAB\u002FMEX implementation not fully optimized (Table I caption)",[41,42,43],"rotating SICK LMS 291 2D laser, one revolution every 2 s, spin axis pitched 65 deg from horizontal (Sec. 3.1)","MicroStrain 3DM-GX2 industrial-grade MEMS IMU on the non-spinning part of the mount (Sec. 3.1)","two fixed vertical SICK LMS 291 lasers in a pushbroom configuration, used only for surface reconstruction, not for trajectory estimation (Sec. 3.1, 4.5)",[45],"vehicle: steel-frame sensor cart strapped to the bed of a site utility vehicle (pickup), driven by a mine employee at 20-30 km\u002Fh under a 30 km\u002Fh limit (Sec. 3.1)","continuous-time non-rigid registration: baseline trajectory plus low-bandwidth corrections parameterized as uniform B-splines (first-order, 0.1 s knots, 4 s window with 2 s shift for the open-loop stage; cubic, 5 s knots for whole-trajectory registration); stacked linearized constraints (surfel match, fixed-surfel, IMU accelerometer and gyro deviation, reference-velocity deviation, smoothness, initial conditions) solved by iteratively reweighted least squares with Cauchy weights and sparse Cholesky factorization (Sec. 2.1, 4.1, 4.2)","surfels from first and second moments in a multi-resolution voxel grid (0.4, 0.8, 1.6, 3.2 m; max 0.5 s time span; at least 15 points and two scans); approximate k-nearest-neighbour search (four neighbours at least 0.5 s apart, modified libnabo) in a weighted position-normal space, with removal of distant and non-reciprocal matches (Sec. 2.1.3, 4.1)","continuous-time: trajectory stored at about 100 Hz samples with spline interpolation; corrections as B-splines (piecewise linear at 0.1 s knots in the sliding window; cubic at 5 s knots globally) (Sec. 2.1.2, 4.1, 4.2)","implicit in continuous-time non-rigid registration; surfel time spans capped (0.5 s in the open-loop stage) to limit uncorrected distortion (Sec. 4.1)","place recognition by keypoint voting: 40 cm downsampled cloud, 20% random keypoints (planar ones discarded), 3D gestalt descriptors over 4 m reduced from 66 to 10 dimensions, places of 3,500 keypoints, RANSAC geometric verification; no false positives on this data set; the mine has no topological loops, so closures come only from backtracking (Sec. 2.2, 4.2)","robust pose-graph optimization (annealed Cauchy M-estimator; rotations solved before translations) for coarse alignment, then whole-trajectory non-rigid registration (cubic B-spline corrections at 5 s knots; surfels with up to 10 s span; five neighbours at least 10 s apart; IMU, smoothness and velocity-deviation constraints); finally registration to the mine survey: 2D gestalt place recognition with 200 m keypoints and 50 m places, pose graph with star-configured survey edges and annealed place-recognition weights, then non-rigid registration against 35,612 fixed surfels built from 17,942 survey points (Sec. 2.2.1, 4.2, 4.3)","surfels for registration; output 3D point cloud (87 million georeferenced points) and a triangulated surface mesh from the vertical lasers, decimated to one-eighth resolution for the user (Sec. 4.5, 6)","no external positioning during acquisition; the final stage uses a pre-existing mine survey (sparse floor-level profile at about 1 m spacing, apparatus and date unknown to the authors) as fixed surfels for georeferencing and drift correction (Sec. 4.3)","6-DoF trajectory (open-loop, closed-loop and survey-registered), georeferenced 3D point cloud and triangulated surface model (Sec. 4, 6)","53.9 min from raw data to the survey-registered model on a 2012 MacBook Pro (2.6 GHz Intel i7), MATLAB\u002FMEX, against 113 min acquisition: open loop 34.8, loop-closure place recognition 4.0, global registration 9.0 (3.9 min surfels plus 5.1 min optimisation), survey place recognition 2.0, survey registration 4.1 min (Table I, Sec. 4.2); about 47.7 min (42% of acquisition) to the closed-loop solution; stationary periods (40% of the data) slow the open-loop stage; 5 Hz open-loop updates reported achievable in real time, with a C++ version under development (Sec. 4.1)",null,"not_verified",[59],{"relation":60,"title":61,"doi_or_url":62},"conference_version","Efficient Large-Scale 3D Mobile Mapping and Surface Reconstruction of an Underground Mine (Field and Service Robotics 2012, Matsushima; Springer Tracts in Advanced Robotics, pp. 479-493, online 2013-12-31). The journal paper states that it extends this publication on the same April 2011 Northparkes deployment, replacing manually extracted survey anchor points with automatic place recognition and adding robust pose-graph optimization, a laser calibration model and tight IMU constraints (JFR Sec. 1, 4.2, 5). LOAM ref. [3] cites this conference version. Conference text itself not read.","10.1007\u002F978-3-642-40686-7_32",{"id":5,"kind":64,"shortName":7,"title":8,"authors":65,"year":9,"venue":68,"venueType":69,"publisher":70,"volumeIssuePages":71,"doi":72,"arxivId":56,"url":73,"firstPublicDate":74,"publicationStatus":16,"metadataStatus":75,"fulltextStatus":15,"era":10,"classicReason":76,"codeUrl":56,"cluster":11,"topics":77,"mdpi":78,"verification":79,"label":6,"fulltextRoute":80,"versionRead":81,"addedByCensus":78},"method",[66,67],"Robert Zlot","Michael Bosse","Journal of Field Robotics","journal","Wiley","31(5): 758-779","10.1002\u002Frob.21504","https:\u002F\u002Fonlinelibrary.wiley.com\u002Fdoi\u002Ffull\u002F10.1002\u002Frob.21504","2014-03-05","metadata_verified","necessary technical node and evaluation precedent: kilometre-scale continuous-time lidar-inertial SLAM in an operating underground mine (Northparkes, 17.1 km) with automatic place recognition, robust pose-graph optimization and registration to a mine survey; it documents that its non-rigid registration originated in bosse_zlot2009_ctscan. The earlier FSR 2012 conference version on the same data set is the version LOAM cites; this journal version is the fuller, automated account and is not the earliest peer-reviewed report of the deployment.",[11],false,"corrected","NTU institutional (Chrome)","version of record, Wiley Online Library full-text HTML (JFR 31(5): 758-779, VoR online 2014-03-05), read in full; equations not rendered in the HTML text",[83,89,94,98,103,109],{"category":84,"model":85,"canonical":85,"role":86,"dataset":56,"specs":87,"locator":88},"lidar","SICK LMS 291 (rotating)","method input","2D laser on a spinning mount, one revolution every 2 s giving a hemispherical 3D view each second; spin axis pitched 65 deg from horizontal; custom driver; internal mirror and encoder wobble calibrated","Sec. 3.1, 3.2, Fig. 1",{"category":90,"model":91,"canonical":91,"role":86,"dataset":56,"specs":92,"locator":93},"imu","MicroStrain 3DM-GX2","industrial-grade MEMS IMU fixed to the non-spinning part of the mount; raw rates used with own bias estimates; accelerometer biases not corrected","Sec. 3.1, 4.1, 4.4",{"category":84,"model":95,"canonical":95,"role":86,"dataset":56,"specs":96,"locator":97},"SICK LMS 291 (two fixed, vertical)","mounted back to back with vertical scan planes covering 360 deg (pushbroom); 160 mm blind spot; used only for surface reconstruction, not for trajectory estimation","Sec. 3.1, 4.5, Fig. 2",{"category":99,"model":100,"canonical":100,"role":86,"dataset":56,"specs":101,"locator":102},"platform","site utility vehicle (pickup truck) carrying a steel-frame sensor cart","cart strapped to the vehicle bed with batteries, electronics and a ROS logging laptop; driven by a mine employee at 20-30 km\u002Fh (limit 30 km\u002Fh)","Sec. 3.1, Fig. 1, Fig. 3",{"category":104,"model":105,"canonical":105,"role":106,"dataset":56,"specs":107,"locator":108},"compute","MacBook Pro (2012), 2.6 GHz Intel i7","compute for runtime","MATLAB\u002FMEX processing, not fully optimised","Sec. 4.1, Table I",{"category":110,"model":111,"canonical":111,"role":112,"dataset":56,"specs":113,"locator":114},"other","mine survey profile (apparatus not reported)","reference or ground truth","17,942 points near floor level at about 1 m spacing, date and method unknown; converted to 35,612 surfels and used as fixed registration targets","Sec. 4.3",[],{"totalRows":117,"groupCount":118,"groups":119,"others":231},10,3,[120,174,204],{"slug":121,"group":122,"sourceId":5,"sourceLabel":6,"table":123,"selfRows":124,"metrics":125,"seqs":131,"entrants":136,"cells":150,"outcomes":167,"locators":168,"hardware":169,"wordings":171,"notes":172},"zlot-bosse2014-mine-table-i","zlot_bosse2014_mine:Table I","Table I",6,[126],{"label":127,"unit":128,"statistic":129,"alignment":130},"computation time (min)","min","not_reported","none",[132],{"dataset":133,"sequence":134,"environment":135},"Northparkes Mine deployment, April 2011 (self-collected)","single run, 17.1 km in 113 min","operating underground copper and gold mine (decline and drives)",[137,140,142,144,146,148],{"name":138,"methodId":5,"linkable":139,"proposed":139,"self":139},"Open-loop trajectory generation (non-rigid registration)",true,{"name":141,"methodId":5,"linkable":139,"proposed":139,"self":139},"Loop closure and coarse global alignment (place recognition)",{"name":143,"methodId":5,"linkable":139,"proposed":139,"self":139},"Global trajectory registration (non-rigid registration)",{"name":145,"methodId":5,"linkable":139,"proposed":139,"self":139},"Coarse alignment to mine survey data (place recognition)",{"name":147,"methodId":5,"linkable":139,"proposed":139,"self":139},"Global registration to mine survey data (non-rigid registration)",{"name":149,"methodId":5,"linkable":139,"proposed":139,"self":139},"Total pipeline",[151,155,158,161,162,164],[152,152,152,153,154,152,152,154,152],0,34.8,-1,[156,152,152,157,154,152,152,154,152],1,4,[159,152,152,160,154,152,152,154,152],2,9,[118,152,152,159,154,152,152,154,152],[157,152,152,163,154,152,152,154,152],4.1,[165,152,152,166,154,152,152,154,152],5,53.9,[],[123],[170],"2012 MacBook Pro, 2.6 GHz Intel i7; MATLAB\u002FMEX, not fully optimised",[],[173],"Computation time per processing stage from raw data to the survey-registered model; acquisition time 113 min",{"slug":175,"group":176,"sourceId":5,"sourceLabel":6,"table":177,"selfRows":159,"metrics":178,"seqs":187,"entrants":189,"cells":192,"outcomes":197,"locators":198,"hardware":200,"wordings":201,"notes":202},"zlot-bosse2014-mine-text-sec-4-3","zlot_bosse2014_mine:Text Sec.4.3","Text Sec.4.3",[179,184],{"label":180,"unit":181,"statistic":182,"alignment":183},"RMS of match error between map surfels and mine-survey surfels along the normal","cm","RMSE","control points",{"label":185,"unit":186,"statistic":129,"alignment":183},"share of map surfels within 50 cm of the corresponding mine-survey surfel","%",[188],{"dataset":133,"sequence":134,"environment":135},[190],{"name":191,"methodId":5,"linkable":139,"proposed":139,"self":139},"Survey-registered solution",[193,195],[152,152,152,194,154,152,154,154,152],25,[152,156,152,196,154,152,154,154,152],95,[],[199],"Sec. 4.3, Fig. 11",[],[],[203],"Map surfels against 35,612 surfels built from the mine survey after the point cloud was registered to that same survey (fixed surfels); a post-registration residual, not an independent accuracy check",{"slug":205,"group":206,"sourceId":5,"sourceLabel":6,"table":207,"selfRows":159,"metrics":208,"seqs":214,"entrants":216,"cells":219,"outcomes":224,"locators":225,"hardware":227,"wordings":228,"notes":229},"zlot-bosse2014-mine-text-sec-4-4","zlot_bosse2014_mine:Text Sec.4.4","Text Sec.4.4",[209,212],{"label":210,"unit":186,"statistic":129,"alignment":211},"open-loop translational drift bias, forward direction (share of distance travelled)","first-pose",{"label":213,"unit":186,"statistic":129,"alignment":211},"open-loop translational drift bias, vertical direction (share of distance travelled)",[215],{"dataset":133,"sequence":134,"environment":135},[217],{"name":218,"methodId":5,"linkable":139,"proposed":139,"self":139},"Open-loop trajectory (non-rigid registration with IMU)",[220,222],[152,152,152,221,154,152,154,154,152],0.2,[152,156,152,223,154,152,154,154,152],0.1,[],[226],"Sec. 4.4, Fig. 10",[],[],[230],"Open-loop drift from segment-wise comparison with the survey-registered trajectory (segments of 10-150 m aligned at the same start time); bias values stated in the text, not read from Fig. 10",[],1790510662590]