[{"data":1,"prerenderedAt":422},["ShallowReactive",2],{"method-demo2014":3},{"method":4,"reference":59,"equipment":81,"figures":117,"results":118},{"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":33,"platform":36,"estimator":39,"association":40,"timeModel":41,"deskew":42,"loopClosure":43,"globalOptimization":44,"mapRepresentation":45,"prior":46,"outputGeometry":47,"compute":48,"codeUrl":49,"codeLicense":50,"relatedVersions":51},"demo2014","Zhang et al., 2014","DEMO","Real-time depth enhanced monocular odometry",2014,"classic","C07","odometry","DEMO 以單眼相機為主，從 RGB-D 相機或 LiDAR 取得深度：先用估測的運動把深度點登錄成局部深度地圖並存進以兩個角度座標建立的 2D KD 樹，再以最近三點構成的小平面內插特徵深度；沒有深度的特徵改用前幾影格的運動三角化，仍無法取得時也保留並以較弱的約束參與求解。逐影格運動用具強健權重的 Levenberg-Marquardt 求解，另以 iSAM 對 8 張影像做低頻光束法平差，最後整合成高頻輸出。","Monocular visual odometry that attaches depth from an RGB-D camera or a LiDAR through a motion-registered depth map, uses features both with and without depth in a robust frame-to-frame solve, and refines motion with low-rate windowed bundle adjustment (iSAM).","full_text_reviewed","peer_reviewed_published","background","DEMO 是早期把 LiDAR 或 RGB-D 深度接到單眼視覺里程計的代表方法，後續視覺 LiDAR 里程計（如 SDV-LOAM）仍以它作為比較基準。論文測試包含室內房間、大廳與戶外道路、草地，沒有施工現場或點雲精度評估；對工地而言，其價值主要在說明深度覆蓋率下降時如何維持追蹤（推論）。",[20,21],"public_benchmark","controlled_experiment",[23,24,25,26,27],"Relative error at trajectory end 1.53 to 3.72 % (RGB-D) and 0.79 to 2.06 % (LiDAR) in four author tests, while Fovis and DVO degrade strongly as depth coverage shrinks in open scenes (Table I)","KITTI training sequences 00-10 mean relative position error 0.93 to 1.87 % (Table II)","Rated second on the KITTI odometry benchmark irrespective of sensing modality and first among visual odometry methods at the time of writing (Abstract)","Bundle adjustment reduced mean error by 0.3 to 0.7 % on KITTI (Sec. VII)","Still uses features without depth, so it keeps working when depth covers only part of the image (Sec. V; Fig. 6)",[29,30,31,32],"Feature tracking (Harris plus KLT) is unreliable in homogeneously coloured corridors (Sec. VIII)","Only RGB-D and LiDAR depth were tested; stereo depth is future work (Sec. VIII)","No loop closure; drift accumulates with distance","Bundle adjustment is less effective on highway scenes, partly due to lower feature quality (Sec. VII)",[34,35],"monocular camera","depth from an RGB-D camera (Xtion Pro Live) or from a 3D LiDAR (rotating Hokuyo UTM-30LX; Velodyne on KITTI)",[37,38],"handheld or mobile sensor rigs (author-collected room, lobby, road and lawn tests)","KITTI vehicle","frame-to-frame motion by Levenberg-Marquardt robust fitting with bisquare weights, using features with depth (two equations each) and without depth (one equation each); sliding bundle adjustment with iSAM on 8 images (one of every five of 40 frames) at about 0.25 to 1.0 Hz; transform integration combines the high-rate and low-rate estimates (Secs. V-VI)","Harris corners tracked by KLT; depth map registered with the estimated motion, downsampled by angular interval and stored in a 2D KD-tree over two angular coordinates; feature depth interpolated from three nearest depth points (planar patch) or, if unavailable, triangulated from previous motion (Sec. V)","discrete image frames (30 Hz in the author tests; 10 Hz on KITTI); depth points are registered into the depth map using the estimated camera motion","not described; LiDAR points are accumulated into the depth map using the estimated motion","no","none beyond the windowed bundle adjustment","local registered depth map (point cloud) around the camera used only for depth association","pre-calibrated camera intrinsics and camera-depth sensor extrinsics (Sec. III)","camera trajectory; registered point clouds of the depth sensor (Fig. 7 shows maps built on KITTI)","laptop with 2.5 GHz cores and 6 GB memory, about three cores used; real-time at the camera rate (Sec. VII)","http:\u002F\u002Fwiki.ros.org\u002Fdemo_rgbd","not verified (ROS wiki pages behind an anti-bot challenge; not bypassed)",[52,56],{"relation":53,"title":54,"doi_or_url":55},"successor","A real-time method for depth enhanced visual odometry (Autonomous Robots 41(1):31-43, 2017; online 2015-12-12; not read)","https:\u002F\u002Fdoi.org\u002F10.1007\u002Fs10514-015-9525-1",{"relation":57,"title":58,"doi_or_url":49},"code_release","demo_rgbd and demo_lidar ROS packages named in the paper (pages behind an anti-bot challenge; availability not verified)",{"id":5,"kind":60,"shortName":7,"title":8,"authors":61,"year":9,"venue":65,"venueType":66,"publisher":67,"volumeIssuePages":68,"doi":69,"arxivId":70,"url":71,"firstPublicDate":72,"publicationStatus":16,"metadataStatus":73,"fulltextStatus":15,"era":10,"classicReason":74,"codeUrl":49,"cluster":11,"topics":75,"mdpi":76,"verification":77,"label":6,"fulltextRoute":78,"versionRead":79,"addedByCensus":80},"method",[62,63,64],"Ji Zhang","Michael Kaess","Sanjiv Singh","2014 IEEE\u002FRSJ International Conference on Intelligent Robots and Systems (IROS)","conference","IEEE","pp. 4973-4980","10.1109\u002Firos.2014.6943269",null,"https:\u002F\u002Fdoi.org\u002F10.1109\u002FIROS.2014.6943269","2014-09","metadata_verified","reproducible baseline and principle reused: first public 2014-09, before the recent window; associates sparse LiDAR or RGB-D depth to monocular features through a registered depth map, mixes features with and without depth in one motion solve, and refines with windowed bundle adjustment; it is the DEMO baseline reported in later visual-LiDAR odometry tables (e.g., SDV-LOAM Table V).",[11],false,"corrected","author copy (institutional repository)","Author copy of the IROS 2014 paper (CMU Robotics Institute pub_files PDF with proceedings page numbers 4973-4980)",true,[82,89,94,99,104,106,112],{"category":83,"model":84,"canonical":84,"role":85,"dataset":86,"specs":87,"locator":88},"rgbd","Xtion Pro Live","method input","DEMO author-collected tests","RGB and depth at 30 Hz, 640 x 480, 58 deg horizontal FoV","Sec. III; Fig. 2(a)",{"category":90,"model":91,"canonical":91,"role":85,"dataset":86,"specs":92,"locator":93},"camera","custom-built camera","up to 60 Hz, 744 x 480, 83 deg horizontal FoV","Sec. III; Fig. 2(b)",{"category":95,"model":96,"canonical":97,"role":85,"dataset":86,"specs":98,"locator":93},"lidar","UTM-30LX (motor-rotated)","Hokuyo UTM-30LX","180 deg FoV, 0.25 deg resolution, 40 lines\u002Fs; rotated by a motor for 3D scanning",{"category":90,"model":100,"canonical":100,"role":85,"dataset":101,"specs":102,"locator":103},"KITTI left monochrome camera","KITTI odometry","10 Hz logging","Sec. VII",{"category":95,"model":105,"canonical":105,"role":85,"dataset":101,"specs":102,"locator":103},"KITTI 360 deg Velodyne laser scanner",{"category":107,"model":108,"canonical":109,"role":110,"dataset":101,"specs":111,"locator":103},"gnss","KITTI high-accuracy GPS\u002FINS","KITTI high accuracy GPS\u002FINS","reference or ground truth","ground truth for KITTI",{"category":113,"model":114,"canonical":114,"role":115,"dataset":70,"specs":116,"locator":103},"compute","laptop","compute for runtime","2.5 GHz cores, 6 GB memory, about three cores used",[],{"totalRows":119,"groupCount":120,"groups":121,"others":421},32,3,[122,295,352],{"slug":123,"group":124,"sourceId":125,"sourceLabel":126,"table":127,"selfRows":128,"metrics":129,"seqs":135,"entrants":163,"cells":174,"outcomes":286,"locators":289,"hardware":291,"wordings":292,"notes":293},"sdvloam2023-table-v","sdvloam2023:Table V","sdvloam2023","Yuan et al., 2023a","Table V",13,[130],{"label":131,"unit":132,"statistic":133,"alignment":134},"Relative translational error (RTE)","%","mean","not_applicable",[136,139,141,143,145,147,149,151,153,155,157,159,161],{"dataset":101,"sequence":137,"environment":138},"00","urban, highway and country driving",{"dataset":101,"sequence":140,"environment":138},"01",{"dataset":101,"sequence":142,"environment":138},"02",{"dataset":101,"sequence":144,"environment":138},"03",{"dataset":101,"sequence":146,"environment":138},"04",{"dataset":101,"sequence":148,"environment":138},"05",{"dataset":101,"sequence":150,"environment":138},"06",{"dataset":101,"sequence":152,"environment":138},"07",{"dataset":101,"sequence":154,"environment":138},"08",{"dataset":101,"sequence":156,"environment":138},"09",{"dataset":101,"sequence":158,"environment":138},"10",{"dataset":101,"sequence":160,"environment":138},"00-10 average",{"dataset":101,"sequence":162,"environment":138},"11-21 mean (KITTI test set)",[164,165,168,170,172],{"name":7,"methodId":5,"linkable":80,"proposed":76,"self":80},{"name":166,"methodId":167,"linkable":80,"proposed":76,"self":76},"LIMO*","limo2018",{"name":169,"methodId":70,"linkable":76,"proposed":76,"self":76},"Huang et al.",{"name":171,"methodId":70,"linkable":76,"proposed":76,"self":76},"DVL-SLAM",{"name":173,"methodId":125,"linkable":80,"proposed":80,"self":76},"Our VO module",[175,179,182,185,187,190,193,196,199,202,205,208,210,212,214,216,218,220,221,222,223,225,227,229,231,233,235,237,239,241,243,244,246,248,250,252,254,255,257,259,261,263,265,266,267,269,271,273,274,276,278,280,281,283,284],[176,176,176,177,178,176,178,178,176],0,1.05,-1,[176,176,180,181,178,176,178,178,176],1,1.87,[176,176,183,184,178,176,178,178,176],2,0.93,[176,176,120,186,178,176,178,178,176],0.99,[176,176,188,189,178,176,178,178,176],4,1.23,[176,176,191,192,178,176,178,178,176],5,1.04,[176,176,194,195,178,176,178,178,176],6,0.96,[176,176,197,198,178,176,178,178,176],7,1.16,[176,176,200,201,178,176,178,178,176],8,1.24,[176,176,203,204,178,176,178,178,176],9,1.17,[176,176,206,207,178,176,178,178,176],10,1.14,[176,176,209,198,178,176,178,178,176],11,[176,176,211,207,178,176,178,178,176],12,[180,176,176,213,178,176,178,178,176],1.12,[180,176,180,215,178,176,178,178,176],0.91,[180,176,188,217,178,176,178,178,176],0.53,[180,176,209,219,176,176,178,178,176],0.85,[180,176,211,184,178,176,178,178,176],[183,176,176,186,178,176,178,178,176],[183,176,180,181,178,176,178,178,176],[183,176,183,224,178,176,178,178,176],1.38,[183,176,120,226,178,176,178,178,176],0.65,[183,176,188,228,178,176,178,178,176],0.42,[183,176,191,230,178,176,178,178,176],0.72,[183,176,194,232,178,176,178,178,176],0.61,[183,176,197,234,178,176,178,178,176],0.56,[183,176,200,236,178,176,178,178,176],1.27,[183,176,203,238,178,176,178,178,176],1.06,[183,176,206,240,178,176,178,178,176],0.83,[183,176,209,242,178,176,178,178,176],0.94,[120,176,176,184,178,176,178,178,176],[120,176,180,245,178,176,178,178,176],1.47,[120,176,183,247,178,176,178,178,176],1.11,[120,176,120,249,178,176,178,178,176],0.92,[120,176,188,251,178,176,178,178,176],0.67,[120,176,191,253,178,176,178,178,176],0.82,[120,176,194,249,178,176,178,178,176],[120,176,197,256,178,176,178,178,176],1.26,[120,176,200,258,178,176,178,178,176],1.32,[120,176,203,260,178,176,178,178,176],0.66,[120,176,206,262,178,176,178,178,176],0.7,[120,176,209,264,180,176,178,178,176],1.98,[188,176,176,251,178,176,178,178,176],[188,176,180,195,178,176,178,178,176],[188,176,183,268,178,176,178,178,176],0.75,[188,176,120,270,178,176,178,178,176],0.86,[188,176,188,272,178,176,178,178,176],0.77,[188,176,191,260,178,176,178,178,176],[188,176,194,275,178,176,178,178,176],0.44,[188,176,197,277,178,176,178,178,176],0.74,[188,176,200,279,178,176,178,178,176],1.07,[188,176,203,217,178,176,178,178,176],[188,176,206,282,178,176,178,178,176],0.51,[188,176,209,230,178,176,178,178,176],[188,176,211,285,178,176,178,178,176],0.88,[287,288],"average over the sequences reported for LIMO* (00, 01, 04) as printed","as printed; the listed 00-10 values average about 0.98",[290],"Table V; Sec. VII",[],[],[294],"KITTI odometry; relative translational error (%) of LiDAR-assisted depth-enhanced visual odometry; baseline values from the original publications; LIMO* uses semantic information; '-' cells not stored",{"slug":296,"group":297,"sourceId":5,"sourceLabel":6,"table":298,"selfRows":209,"metrics":299,"seqs":303,"entrants":330,"cells":333,"outcomes":345,"locators":346,"hardware":348,"wordings":349,"notes":350},"demo2014-table-ii","demo2014:Table II","Table II",[300],{"label":301,"unit":132,"statistic":133,"alignment":302},"Mean relative position error","not_reported",[304,307,310,313,316,318,320,322,324,326,328],{"dataset":101,"sequence":305,"environment":306},"00 (3714 m)","Urban",{"dataset":101,"sequence":308,"environment":309},"01 (4268 m)","Highway",{"dataset":101,"sequence":311,"environment":312},"02 (5075 m)","Urban + Country",{"dataset":101,"sequence":314,"environment":315},"03 (563 m)","Country",{"dataset":101,"sequence":317,"environment":315},"04 (397 m)",{"dataset":101,"sequence":319,"environment":306},"05 (2223 m)",{"dataset":101,"sequence":321,"environment":306},"06 (1239 m)",{"dataset":101,"sequence":323,"environment":306},"07 (695 m)",{"dataset":101,"sequence":325,"environment":312},"08 (3225 m)",{"dataset":101,"sequence":327,"environment":312},"09 (1717 m)",{"dataset":101,"sequence":329,"environment":312},"10 (919 m)",[331],{"name":332,"methodId":5,"linkable":80,"proposed":80,"self":80},"DEMO (camera + LiDAR)",[334,335,336,337,338,339,340,341,342,343,344],[176,176,176,177,178,176,178,178,176],[176,176,180,181,178,176,178,178,176],[176,176,183,184,178,176,178,178,176],[176,176,120,186,178,176,178,178,176],[176,176,188,189,178,176,178,178,176],[176,176,191,192,178,176,178,178,176],[176,176,194,195,178,176,178,178,176],[176,176,197,198,178,176,178,178,176],[176,176,200,201,178,176,178,178,176],[176,176,203,204,178,176,178,178,176],[176,176,206,207,178,176,178,178,176],[],[347],"Table II; Sec. VII",[],[],[351],"KITTI odometry training sequences 00-10; left monochrome camera plus Velodyne; mean relative position error as a percentage (KITTI protocol)",{"slug":353,"group":354,"sourceId":5,"sourceLabel":6,"table":355,"selfRows":200,"metrics":356,"seqs":359,"entrants":372,"cells":382,"outcomes":414,"locators":415,"hardware":417,"wordings":418,"notes":419},"demo2014-table-i","demo2014:Table I","Table I",[357],{"label":358,"unit":132,"statistic":302,"alignment":302},"Relative position error (% of distance traveled)",[360,363,366,369],{"dataset":86,"sequence":361,"environment":362},"Room (16 m)","indoor conference room",{"dataset":86,"sequence":364,"environment":365},"Lobby (56 m)","indoor lobby",{"dataset":86,"sequence":367,"environment":368},"Road (87 m)","outdoor road",{"dataset":86,"sequence":370,"environment":371},"Lawn (86 m)","outdoor lawn",[373,376,378,380],{"name":374,"methodId":375,"linkable":80,"proposed":76,"self":76},"Fovis","fovis2017",{"name":377,"methodId":70,"linkable":76,"proposed":76,"self":76},"DVO",{"name":379,"methodId":5,"linkable":80,"proposed":80,"self":80},"Our VO (RGB-D)",{"name":381,"methodId":5,"linkable":80,"proposed":80,"self":80},"Our VO (Lidar)",[383,385,387,389,391,392,394,396,398,400,402,404,406,408,410,412],[176,176,176,384,178,176,178,178,176],2.72,[176,176,180,386,178,176,178,178,176],5.56,[176,176,183,388,178,176,178,178,176],13.04,[176,176,120,390,178,176,178,178,176],9.97,[180,176,176,181,178,176,178,178,176],[180,176,180,393,178,176,178,178,176],8.36,[180,176,183,395,178,176,178,178,176],13.6,[180,176,120,397,178,176,178,178,176],32.07,[183,176,176,399,178,176,178,178,176],2.14,[183,176,180,401,178,176,178,178,176],1.84,[183,176,183,403,178,176,178,178,176],1.53,[183,176,120,405,178,176,178,178,176],3.72,[120,176,176,407,178,176,178,178,176],2.06,[120,176,180,409,178,176,178,178,176],1.79,[120,176,183,411,178,176,178,178,176],0.79,[120,176,120,413,178,176,178,178,176],1.73,[],[416],"Table I; Sec. VII",[],[],[420],"Author-collected tests with the Xtion RGB-D camera and with the custom camera plus rotating Hokuyo LiDAR; the camera starts and stops at the same position and the gap between the trajectory ends divided by trajectory length is the relative position error (3D coordinates); Fovis and DVO use RGB-D input",[],1790510657838]