[{"data":1,"prerenderedAt":705},["ShallowReactive",2],{"method-fovis2017":3},{"method":4,"reference":60,"equipment":86,"figures":123,"results":124},{"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":23,"limitations":29,"sensors":35,"platform":38,"estimator":40,"association":41,"timeModel":42,"deskew":43,"loopClosure":44,"globalOptimization":45,"mapRepresentation":46,"prior":47,"outputGeometry":48,"compute":49,"codeUrl":50,"codeLicense":51,"relatedVersions":52},"fovis2017","Huang et al., 2017","FOVIS","Visual Odometry and Mapping for Autonomous Flight Using an RGB-D Camera",2017,"classic","C08","odometry","本章提出供四旋翼無人機自主飛行使用的 RGB-D 視覺里程計，後來以 fovis 函式庫公開。演算法沿用立體視覺里程計的標準流程：灰階影像建立三層高斯金字塔，以自適應門檻的 FAST 角點擷取特徵並分格保留，從深度影像取得特徵深度；先以縮小影像直接估計初始旋轉，藉此限縮搜尋視窗，再以 9×9 影像塊描述子做雙向一致的匹配與次像素修正。內點以「剛體運動保持點間距離」建立一致性圖，並以貪婪法近似最大團挑選；位姿先以 Horn 絕對定向求解，再最小化重投影誤差，並以參考關鍵影格降低懸停時的漂移。里程計與 IMU 以 EKF 融合後在機上即時控制飛行；迴圈閉合與位姿圖最佳化則沿用作者先前的 RGB-D Mapping，在機外筆電執行，並建立 10 公分解析度的佔據體素地圖。","Feature-based RGB-D visual odometry for onboard MAV control, released as the fovis library: FAST features on a Gaussian pyramid with bucketing, an image-based initial rotation to constrain matching, SAD patch matching with sub-pixel ESM refinement, greedy max-clique inlier detection and reprojection-error refinement against a reference keyframe; loop closure (RGB-D Mapping, TORO) and a 10 cm occupancy map run offboard.","full_text_reviewed","peer_reviewed_published","background","論文未在施工現場測試；實驗在約 11 × 7 × 4 公尺的動作捕捉室，以及 MIT 校園與 Intel 西雅圖辦公室的室內空間進行（Sec. 4）。fovis 後來成為 Kintinuous 的里程計前端，出現在 ICL-NUIM 的軌跡評估中 [handa2014iclnuim]，也是 RTAB-Map 比較的里程計選項之一 [rtabmap2019]。作者指出特徵少的區域與超出 Kinect 量測距離的大空間會使其表現變差（Sec. 5），因此在白牆多、樓板空曠的施工中建築內單獨使用時容易失效，需要其他感測器輔助（推論）。",[20,21,22],"controlled_experiment","independent_reference","completed_building",[24,25,26,27,28],"On a deliberately challenging motion-capture dataset the chosen configuration had 0.387 m\u002Fs mean velocity error and 3.39% gross failures at 14.7 ms per frame; in feature-rich environments the authors observed 0.08 m\u002Fs mean velocity error without gross failures (Sec. 4.1; Table 1)","Greedy max-clique inlier detection outperformed RANSAC and preemptive RANSAC at comparable computation (Sec. 4.1; Table 1)","Visual odometry, sensor fusion and control run onboard at roughly 25 ms per frame on the 1.86 GHz computer (Sec. 4.1; Sec. 5)","Under visual-odometry control the vehicle held position for 90 s with 6.2 cm mean and 19 cm maximum deviation measured by motion capture (Fig. 4)","Keyframe reference frames eliminate drift when the viewpoint does not change much, which helps while hovering (Sec. 3.1)",[30,31,32,33,34],"Performs poorly in regions with few visual features and in large open areas where structure lies beyond the Kinect range; works better in cluttered, close quarters (Sec. 5)","Assumes a static environment and slow vehicle motion; faster flight brings motion blur and Kinect rolling-shutter artifacts (Sec. 5)","With moving objects covering much of the image, the maximal clique may not correspond to the static scene (Sec. 5)","Loop closing and SLAM are not fast enough to run on the onboard processor (Sec. 3.2; Sec. 5)","Maps from the larger environments have no ground truth; their quality is judged visually from the rendered point cloud (Sec. 4.2)",[36,37],"RGB-D camera: stripped-down Microsoft Kinect (PrimeSense), 640 x 480 RGB-D at 30 Hz","IMU on the vehicle, fused with the visual odometry in an EKF for control (not used inside the visual odometry)",[39],"UAV (AscTec Pelican quadrotor; motion capture room of about 11 m x 7 m x 4 m; autonomous flights around the MIT campus and at the Intel Research office in Seattle)","Frame-to-reference-keyframe motion from sparse 3D feature matches: Horn absolute orientation on the inliers, refined by nonlinear least-squares minimization of feature reprojection error (bidirectional, ESM), then refined again after discarding matches above a fixed reprojection threshold; the reference frame is replaced only when motion against it fails or has too few inliers. For flight control the VO is fused with IMU data in an EKF, and delayed SLAM corrections are applied retroactively to the state history.","FAST corners on a three-level Gaussian pyramid with an adaptive threshold and 80 x 80 pixel bucketing (25 strongest per bucket), depth read from the depth image; 80-byte descriptors from 9 x 9 intensity patches matched by sum of absolute differences with a mutual-consistency check inside a search window set by an image-based initial rotation estimate; sub-pixel refinement with ESM; inliers from a greedy approximation of the maximal clique of matches whose 3D distances are preserved","discrete poses (frame to reference keyframe)","not_applicable (RGB-D camera; Kinect rolling shutter named as a limitation at higher speed, Sec. 5)","Not part of the visual odometry. In the paper's full system RGB-D Mapping [14] runs offboard: keyframes every 10 deg or 25 cm, candidates limited to 90 deg and 5 m pose difference and to the 15 best vocabulary-tree matches, RANSAC over FAST keypoints with Calonder descriptors (ratio 0.6, at least 10 inliers), and a two-frame sparse bundle adjustment of the relative pose","Offboard pose graph optimized with TORO after each loop closure (roughly 30 ms); corrected poses and voxel maps sent back to the vehicle; sparse bundle adjustment over all features only offline (Fig. 7)","Offboard 3D log-likelihood occupancy voxel grid at 10 cm resolution from depth downsampled to 128 x 96 (about 1.5 ms per frame); rendered point clouds; offline textured surfaces from sparse bundle adjustment","none","real-time 6-DoF pose and velocity estimates; occupancy voxel map used for path planning; rendered RGB-D point cloud; offline textured surface model","Visual odometry onboard on a 1.86 GHz Core2Duo flight computer with 4 GB RAM (Pixhawk project), roughly 25 ms per frame; 14.7 ms per frame in Table 1 on a laptop (written as 2.67 GHz in Sec. 4.1 and 2.6 GHz in the timing paragraph); loop closure and mapping offboard on a laptop","https:\u002F\u002Fgithub.com\u002Ffovis\u002Ffovis","GPL (repository contains a GPL-2 licence file, while the project page fovis.github.io states GPL version 3 or later; no SPDX licence detected by GitHub)",[53,57],{"relation":54,"title":55,"doi_or_url":56},"conference_version","ISRR 2011 paper, Flagstaff, Arizona, USA, Aug. 2011 (author PDF linked from the fovis project page)","https:\u002F\u002Ffovis.github.io",{"relation":58,"title":59,"doi_or_url":50},"code_release","fovis\u002Ffovis (libfovis visual odometry library)",{"id":5,"kind":61,"shortName":7,"title":8,"authors":62,"year":9,"venue":70,"venueType":71,"publisher":72,"volumeIssuePages":73,"doi":74,"arxivId":75,"url":76,"firstPublicDate":77,"publicationStatus":16,"metadataStatus":78,"fulltextStatus":15,"era":10,"classicReason":79,"codeUrl":50,"cluster":11,"topics":80,"mdpi":81,"verification":82,"label":6,"fulltextRoute":83,"versionRead":84,"addedByCensus":85},"method",[63,64,65,66,67,68,69],"Albert S. Huang","Abraham Bachrach","Peter Henry","Michael Krainin","Daniel Maturana","Dieter Fox","Nicholas Roy","Robotics Research (ISRR 2011), Springer Tracts in Advanced Robotics, vol. 100 (eds. H.I. Christensen and O. Khatib)","book_chapter","Springer International Publishing","STAR 100:235-252","10.1007\u002F978-3-319-29363-9_14",null,"https:\u002F\u002Fapi.crossref.org\u002Fworks\u002F10.1007\u002F978-3-319-29363-9_14","2011-08","metadata_verified","tool still in use: the fovis library implementing this feature-based RGB-D visual odometry is the odometry front end of the Kintinuous rows in the ICL-NUIM benchmark (handa2014iclnuim) and an odometry option evaluated in RTAB-Map (rtabmap2019).",[11],false,"corrected","NTU institutional (curl)","Springer version of record (Robotics Research, STAR vol. 100, pp. 235-252, (c) Springer International Publishing Switzerland 2017, online 2016-08-26); the 2011 ISRR author copy was not compared",true,[87,93,98,103,109,113,117],{"category":88,"model":89,"canonical":89,"role":90,"dataset":75,"specs":91,"locator":92},"rgbd","Microsoft Kinect (stripped down)","method input","640 x 480 RGB-D image at 30 Hz; 115 g when stripped down; mounted at the base of the vehicle tilted slightly down","Sec. 1; Sec. 3; Fig. 1",{"category":94,"model":95,"canonical":95,"role":90,"dataset":75,"specs":96,"locator":97},"imu","onboard IMU (model not reported)","fused with visual-odometry motion estimates in an Extended Kalman Filter for position and velocity","Sec. 1; Sec. 3.3",{"category":99,"model":100,"canonical":100,"role":90,"dataset":75,"specs":101,"locator":102},"platform","Pelican quadrotor","maximal dimension 70 cm, payload up to 1000 g","Sec. 3; Fig. 1",{"category":104,"model":105,"canonical":105,"role":106,"dataset":75,"specs":107,"locator":108},"compute","onboard flight computer (Pixhawk project, ETH Zurich)","compute for runtime","1.86 GHz Core2Duo processor, 4 GB RAM; runs VO, state estimation and control; roughly 25 ms per VO frame","Sec. 3; Sec. 4.1",{"category":104,"model":110,"canonical":110,"role":106,"dataset":75,"specs":111,"locator":112},"laptop computer (2.67 GHz per Sec. 4.1; 2.6 GHz per the timing paragraph; model not reported)","used for the timing results of Table 1","Sec. 4.1",{"category":104,"model":114,"canonical":114,"role":106,"dataset":75,"specs":115,"locator":116},"offboard laptop (model not reported)","receives RGB-D data from the MAV; detects loop closures, computes global pose corrections and builds the occupancy voxel map","Sec. 3.2",{"category":118,"model":119,"canonical":119,"role":120,"dataset":75,"specs":121,"locator":122},"other","motion capture system (model not reported)","reference or ground truth","120 Hz ground truth of MAV position and attitude; room about 11 m x 7 m x 4 m","Sec. 4.1; Fig. 4",[],{"totalRows":125,"groupCount":126,"groups":127,"others":693},94,6,[128,309,572,623],{"slug":129,"group":130,"sourceId":5,"sourceLabel":6,"table":131,"selfRows":132,"metrics":133,"seqs":147,"entrants":152,"cells":189,"outcomes":303,"locators":304,"hardware":305,"wordings":306,"notes":307},"fovis2017-table-1","fovis2017:Table 1","Table 1",54,[134,138,142,145],{"label":135,"unit":136,"statistic":137,"alignment":47},"Velocity error (m\u002Fs), printed with +- 0.004 (undefined in the chapter)","m\u002Fs","mean",{"label":139,"unit":140,"statistic":141,"alignment":47},"% gross failures","%","not_reported",{"label":143,"unit":144,"statistic":141,"alignment":47},"Total time (ms)","ms",{"label":146,"unit":136,"statistic":137,"alignment":47},"Velocity error (m\u002Fs), printed with +- 0.005 (undefined in the chapter)",[148],{"dataset":149,"sequence":150,"environment":151},"authors' MAV motion-capture dataset","MAV flight in motion capture room","indoor motion capture room about 11 m x 7 m x 4 m, one blank wall, quadrotor flight",[153,155,157,159,161,163,165,167,169,171,173,175,177,179,181,183,185,187],{"name":154,"methodId":5,"linkable":85,"proposed":85,"self":85},"Our approach (greedy max-clique, initial rotation, 3 pyramid levels, bidirectional ESM, 9 x 9 window, subpixel refinement, adaptive FAST threshold, grid bucketing)",{"name":156,"methodId":5,"linkable":85,"proposed":81,"self":85},"Inlier detection: RANSAC",{"name":158,"methodId":5,"linkable":85,"proposed":81,"self":85},"Inlier detection: Preemptive RANSAC",{"name":160,"methodId":5,"linkable":85,"proposed":81,"self":85},"Initial rotation estimate: None",{"name":162,"methodId":5,"linkable":85,"proposed":81,"self":85},"Gaussian pyramid levels: 1",{"name":164,"methodId":5,"linkable":85,"proposed":81,"self":85},"Gaussian pyramid levels: 2",{"name":166,"methodId":5,"linkable":85,"proposed":81,"self":85},"Gaussian pyramid levels: 4",{"name":168,"methodId":5,"linkable":85,"proposed":81,"self":85},"Reprojection error minimization: Bidir. Gauss-Newton",{"name":170,"methodId":5,"linkable":85,"proposed":81,"self":85},"Reprojection error minimization: Unidir. Gauss-Newton",{"name":172,"methodId":5,"linkable":85,"proposed":81,"self":85},"Reprojection error minimization: Unidir. ESM",{"name":174,"methodId":5,"linkable":85,"proposed":81,"self":85},"Reprojection error minimization: Absolute orientation only",{"name":176,"methodId":5,"linkable":85,"proposed":81,"self":85},"Feature window size: 3",{"name":178,"methodId":5,"linkable":85,"proposed":81,"self":85},"Feature window size: 5",{"name":180,"methodId":5,"linkable":85,"proposed":81,"self":85},"Feature window size: 7",{"name":182,"methodId":5,"linkable":85,"proposed":81,"self":85},"Feature window size: 11",{"name":184,"methodId":5,"linkable":85,"proposed":81,"self":85},"Subpixel feature refinement: No refinement",{"name":186,"methodId":5,"linkable":85,"proposed":81,"self":85},"Adaptive FAST threshold: Fixed threshold (10)",{"name":188,"methodId":5,"linkable":85,"proposed":81,"self":85},"Feature grid\u002Fbucketing: No grid",[190,194,197,200,203,205,207,209,211,213,215,217,219,221,223,225,228,230,232,233,235,237,239,241,242,245,247,249,251,253,254,257,259,261,263,265,267,269,271,273,275,277,279,281,283,285,288,290,292,294,296,297,299,301],[191,191,191,192,193,191,193,193,191],0,0.387,-1,[191,195,191,196,193,191,193,193,191],1,3.39,[191,198,191,199,193,191,191,193,191],2,14.7,[195,201,191,202,193,191,193,193,191],3,0.412,[195,195,191,204,193,191,193,193,191],6.05,[195,198,191,206,193,191,191,193,191],15.3,[198,201,191,208,193,191,193,193,191],0.414,[198,195,191,210,193,191,193,193,191],5.91,[198,198,191,212,193,191,191,193,191],14.9,[201,191,191,214,193,191,193,193,191],0.388,[201,195,191,216,193,191,193,193,191],4.22,[201,198,191,218,193,191,191,193,191],13.6,[220,191,191,192,193,191,193,193,191],4,[220,195,191,222,193,191,193,193,191],5.17,[220,198,191,224,193,191,191,193,191],17,[226,191,191,227,193,191,193,193,191],5,0.385,[226,195,191,229,193,191,193,193,191],3.52,[226,198,191,231,193,191,191,193,191],15.1,[126,191,191,192,193,191,193,193,191],[126,195,191,234,193,191,193,193,191],3.5,[126,198,191,236,193,191,191,193,191],14.5,[238,191,191,192,193,191,193,193,191],7,[238,195,191,240,193,191,193,193,191],3.24,[238,198,191,199,193,191,191,193,191],[243,191,191,244,193,191,193,193,191],8,0.391,[243,195,191,246,193,191,193,193,191],3.45,[243,198,191,248,193,191,191,193,191],14.6,[250,191,191,244,193,191,193,193,191],9,[250,195,191,252,193,191,193,193,191],3.47,[250,198,191,248,193,191,191,193,191],[255,201,191,256,193,191,193,193,191],10,0.467,[255,195,191,258,193,191,193,193,191],10.97,[255,198,191,260,193,191,191,193,191],14.4,[262,191,191,244,193,191,193,193,191],11,[262,195,191,264,193,191,193,193,191],5.96,[262,198,191,266,193,191,191,193,191],12.8,[268,191,191,214,193,191,193,193,191],12,[268,195,191,270,193,191,193,193,191],4.24,[268,198,191,272,193,191,191,193,191],13.7,[274,191,191,214,193,191,193,193,191],13,[274,195,191,276,193,191,193,193,191],3.72,[274,198,191,278,193,191,191,193,191],14.2,[280,191,191,214,193,191,193,193,191],14,[280,195,191,282,193,191,193,193,191],3.42,[280,198,191,284,193,191,191,193,191],15.7,[286,191,191,287,193,191,193,193,191],15,0.404,[286,195,191,289,193,191,193,193,191],5.13,[286,198,191,291,193,191,191,193,191],13.1,[293,191,191,227,193,191,193,193,191],16,[293,195,191,295,193,191,193,193,191],3.12,[293,198,191,206,193,191,191,193,191],[224,191,191,298,193,191,193,193,191],0.398,[224,195,191,300,193,191,193,193,191],4.02,[224,198,191,302,193,191,191,193,191],24.6,[],[131],[110],[],[308],"Ablation on a challenging MAV motion-capture dataset (motion blur, feature-poor images); each row changes one component of the authors' configuration; velocity error is the mean velocity error magnitude against differentiated motion-capture data; a gross failure is no estimate or velocity error above 1 m\u002Fs; total time per RGB-D frame on a laptop",{"slug":310,"group":311,"sourceId":312,"sourceLabel":313,"table":131,"selfRows":314,"metrics":315,"seqs":320,"entrants":367,"cells":380,"outcomes":566,"locators":567,"hardware":568,"wordings":569,"notes":570},"mrsmap2014-table-1","mrsmap2014:Table 1","mrsmap2014","Stückler & Behnke, 2014",22,[316],{"label":317,"unit":318,"statistic":319,"alignment":141},"median relative pose error (RPE) in mm","mm","median",[321,325,327,329,331,333,335,337,339,341,343,345,347,349,351,353,355,357,359,361,363,365],{"dataset":322,"sequence":323,"environment":324},"TUM RGB-D (Freiburg)","fr1 360","indoor office and structure\u002Ftexture test scenes, RGB-D camera (carrying mode not stated in the paper)",{"dataset":322,"sequence":326,"environment":324},"fr1 desk",{"dataset":322,"sequence":328,"environment":324},"fr1 desk2",{"dataset":322,"sequence":330,"environment":324},"fr1 floor",{"dataset":322,"sequence":332,"environment":324},"fr1 plant",{"dataset":322,"sequence":334,"environment":324},"fr1 room",{"dataset":322,"sequence":336,"environment":324},"fr1 rpy",{"dataset":322,"sequence":338,"environment":324},"fr1 teddy",{"dataset":322,"sequence":340,"environment":324},"fr1 xyz",{"dataset":322,"sequence":342,"environment":324},"fr2 desk",{"dataset":322,"sequence":344,"environment":324},"fr2 large no loop",{"dataset":322,"sequence":346,"environment":324},"fr2 rpy",{"dataset":322,"sequence":348,"environment":324},"fr2 xyz",{"dataset":322,"sequence":350,"environment":324},"fr3 long office household",{"dataset":322,"sequence":352,"environment":324},"fr3 nostruct. notext. far",{"dataset":322,"sequence":354,"environment":324},"fr3 nostruct. notext. near",{"dataset":322,"sequence":356,"environment":324},"fr3 nostruct. text. far",{"dataset":322,"sequence":358,"environment":324},"fr3 nostruct. text. near",{"dataset":322,"sequence":360,"environment":324},"fr3 struct. notext. far",{"dataset":322,"sequence":362,"environment":324},"fr3 struct. notext. near",{"dataset":322,"sequence":364,"environment":324},"fr3 struct. text. far",{"dataset":322,"sequence":366,"environment":324},"fr3 struct. text. near",[368,370,372,375,378],{"name":369,"methodId":312,"linkable":85,"proposed":85,"self":81},"Ours (MRSMap)",{"name":371,"methodId":75,"linkable":81,"proposed":81,"self":81},"Warp [17] (OpenCV)",{"name":373,"methodId":374,"linkable":85,"proposed":81,"self":81},"GICP [5]","segal2009gicp",{"name":376,"methodId":377,"linkable":85,"proposed":81,"self":81},"3D-NDT [7]","stoyanov2012d2dndt",{"name":379,"methodId":5,"linkable":85,"proposed":81,"self":85},"Fovis [12]",[381,383,385,387,389,391,393,395,397,399,401,403,405,407,409,411,413,415,416,417,419,420,422,424,426,428,429,430,431,433,435,436,437,438,440,441,442,443,445,447,448,449,451,453,455,456,457,458,460,462,464,466,468,469,471,472,474,476,478,479,480,482,483,484,485,487,488,490,491,492,494,496,498,500,502,504,506,508,510,512,514,516,518,520,522,524,526,527,529,531,533,536,537,538,540,542,544,545,546,548,550,553,555,556,557,558,560,561,563,564],[191,191,191,382,193,191,193,193,191],5.1,[195,191,191,384,193,191,193,193,191],5.9,[198,191,191,386,193,191,193,193,191],18.8,[201,191,191,388,193,191,193,193,191],7.8,[220,191,191,390,193,191,193,193,191],7.1,[191,191,195,392,193,191,193,193,191],4.4,[195,191,195,394,193,191,193,193,191],5.8,[198,191,195,396,193,191,193,193,191],10.2,[201,191,195,398,193,191,193,193,191],7.9,[220,191,195,400,193,191,193,193,191],6.3,[191,191,198,402,193,191,193,193,191],4.5,[195,191,198,404,193,191,193,193,191],6.2,[198,191,198,406,193,191,193,193,191],10.4,[201,191,198,408,193,191,193,193,191],8.2,[220,191,198,410,193,191,193,193,191],6.6,[191,191,201,412,193,191,193,193,191],4.9,[195,191,201,414,193,191,193,193,191],2.1,[198,191,201,226,193,191,193,193,191],[201,191,201,400,193,191,193,193,191],[220,191,201,418,193,191,193,193,191],2.6,[191,191,220,234,193,191,193,193,191],[195,191,220,421,193,191,193,193,191],4.2,[198,191,220,423,193,191,193,193,191],16.1,[201,191,220,425,193,191,193,193,191],7.4,[220,191,220,427,193,191,193,193,191],4.6,[191,191,226,234,193,191,193,193,191],[195,191,226,427,193,191,193,193,191],[198,191,226,396,193,191,193,193,191],[201,191,226,432,193,191,193,193,191],6.1,[220,191,226,434,193,191,193,193,191],5.4,[191,191,126,201,193,191,193,193,191],[195,191,126,382,193,191,193,193,191],[198,191,126,406,193,191,193,193,191],[201,191,126,439,193,191,193,193,191],6.8,[220,191,126,434,193,191,193,193,191],[191,191,238,421,193,191,193,193,191],[195,191,238,432,193,191,193,193,191],[198,191,238,444,193,191,193,193,191],21.3,[201,191,238,446,193,191,193,193,191],8.8,[220,191,238,390,193,191,193,193,191],[191,191,243,418,193,191,193,193,191],[195,191,243,450,193,191,193,193,191],4.1,[198,191,243,452,193,191,193,193,191],3.9,[201,191,243,454,193,191,193,193,191],5.2,[220,191,243,427,193,191,193,193,191],[191,191,250,414,193,191,193,193,191],[195,191,250,414,193,191,193,193,191],[198,191,250,459,193,191,193,193,191],6.7,[201,191,250,461,193,191,193,193,191],4.3,[220,191,250,463,193,191,193,193,191],2.5,[191,191,255,465,193,191,193,193,191],21.8,[195,191,255,467,193,191,193,193,191],20.5,[198,191,255,444,193,191,193,193,191],[201,191,255,470,193,191,193,193,191],32.1,[220,191,255,262,193,191,193,193,191],[191,191,262,473,193,191,193,193,191],1.6,[195,191,262,475,193,191,193,193,191],1.7,[198,191,262,477,193,191,193,193,191],1.3,[201,191,262,421,193,191,193,193,191],[220,191,262,475,193,191,193,193,191],[191,191,268,481,193,191,193,193,191],1.4,[195,191,268,198,193,191,193,193,191],[198,191,268,475,193,191,193,193,191],[201,191,268,220,193,191,193,193,191],[220,191,268,486,193,191,193,193,191],1.9,[191,191,274,418,193,191,193,193,191],[195,191,274,489,193,191,193,193,191],3.2,[198,191,274,388,193,191,193,193,191],[201,191,274,421,193,191,193,193,191],[220,191,274,493,193,191,193,193,191],3.7,[191,191,280,495,193,191,193,193,191],9.7,[195,191,280,497,193,191,193,193,191],40.4,[198,191,280,499,193,191,193,193,191],8.6,[201,191,280,501,193,191,193,193,191],13.8,[220,191,280,503,193,191,193,193,191],11.3,[191,191,286,505,193,191,193,193,191],15.2,[195,191,286,507,193,191,193,193,191],28.2,[198,191,286,509,193,191,193,193,191],12.5,[201,191,286,511,193,191,193,193,191],17.1,[220,191,286,513,193,191,193,193,191],11.2,[191,191,293,515,193,191,193,193,191],18.5,[195,191,293,517,193,191,193,193,191],19.2,[198,191,293,519,193,191,193,193,191],10.9,[201,191,293,521,193,191,193,193,191],18.6,[220,191,293,523,193,191,193,193,191],20.8,[191,191,224,525,193,191,193,193,191],11.5,[195,191,224,238,193,191,193,193,191],[198,191,224,528,193,191,193,193,191],8.9,[201,191,224,530,193,191,193,193,191],10.6,[220,191,224,532,193,191,193,193,191],7.3,[191,191,534,535,193,191,193,193,191],18,2.2,[195,191,534,499,193,191,193,193,191],[198,191,534,402,193,191,193,193,191],[201,191,534,539,193,191,193,193,191],2.9,[220,191,534,541,193,191,193,193,191],9.1,[191,191,543,414,193,191,193,193,191],19,[195,191,543,499,193,191,193,193,191],[198,191,543,539,193,191,193,193,191],[201,191,543,547,193,191,193,193,191],2.4,[220,191,543,549,193,191,193,193,191],9.3,[191,191,551,552,193,191,193,193,191],20,5.5,[195,191,551,554,193,191,193,193,191],8.1,[198,191,551,390,193,191,193,193,191],[201,191,551,434,193,191,193,193,191],[220,191,551,446,193,191,193,193,191],[191,191,559,489,193,191,193,193,191],21,[195,191,559,384,193,191,193,193,191],[198,191,559,562,193,191,193,193,191],5.6,[201,191,559,552,193,191,193,193,191],[220,191,559,565,193,191,193,193,191],6.5,[],[131],[],[],[571],"Incremental (frame-to-frame) registration on TUM Freiburg sequences; median translational relative pose error in mm (maximum values in brackets in the table not extracted); warp is the OpenCV reimplementation",{"slug":573,"group":574,"sourceId":5,"sourceLabel":6,"table":575,"selfRows":243,"metrics":576,"seqs":594,"entrants":599,"cells":602,"outcomes":615,"locators":616,"hardware":617,"wordings":619,"notes":620},"fovis2017-text-sec-4-1","fovis2017:Text Sec.4.1","Text Sec.4.1",[577,579,581,583,585,587,590,592],{"label":578,"unit":144,"statistic":141,"alignment":47},"Preprocessing time",{"label":580,"unit":144,"statistic":141,"alignment":47},"Feature extraction time",{"label":582,"unit":144,"statistic":141,"alignment":47},"Initial rotation estimation time",{"label":584,"unit":144,"statistic":141,"alignment":47},"Feature matching time",{"label":586,"unit":144,"statistic":141,"alignment":47},"Inlier detection time",{"label":588,"unit":144,"statistic":589,"alignment":47},"Motion estimation time (text: less than 0.1 ms)","max",{"label":591,"unit":144,"statistic":141,"alignment":47},"runtime per frame on the onboard computer (text: roughly 25 ms)",{"label":593,"unit":136,"statistic":137,"alignment":47},"mean velocity error",[595,596],{"dataset":149,"sequence":150,"environment":151},{"dataset":597,"sequence":141,"environment":598},"authors' observations in feature-rich environments","indoor, feature-rich (not specified)",[600],{"name":601,"methodId":5,"linkable":85,"proposed":85,"self":85},"our approach (visual odometry of this chapter; the name 'fovis' is not used in the text)",[603,604,606,607,608,609,611,613],[191,191,191,414,193,191,191,193,191],[191,195,191,605,193,191,191,193,191],3.1,[191,198,191,195,193,191,191,193,191],[191,201,191,126,193,191,191,193,191],[191,220,191,535,193,191,191,193,191],[191,226,191,610,193,191,191,193,191],0.1,[191,126,191,612,193,191,195,193,191],25,[191,238,195,614,193,191,193,193,195],0.08,[],[112],[110,618],"onboard 1.86 GHz Core2Duo flight computer, 4 GB RAM",[],[621,622],"Per-stage timing of the chosen configuration stated in the Timing paragraph (laptop given as 2.6 GHz there), plus the approximate onboard time per frame","Velocity error the authors observed in environments with richer visual features than the benchmark room (no gross failures); no dataset details given",{"slug":624,"group":625,"sourceId":626,"sourceLabel":627,"table":628,"selfRows":220,"metrics":629,"seqs":632,"entrants":646,"cells":655,"outcomes":686,"locators":687,"hardware":689,"wordings":690,"notes":691},"demo2014-table-i","demo2014:Table I","demo2014","Zhang et al., 2014","Table I",[630],{"label":631,"unit":140,"statistic":141,"alignment":141},"Relative position error (% of distance traveled)",[633,637,640,643],{"dataset":634,"sequence":635,"environment":636},"DEMO author-collected tests","Room (16 m)","indoor conference room",{"dataset":634,"sequence":638,"environment":639},"Lobby (56 m)","indoor lobby",{"dataset":634,"sequence":641,"environment":642},"Road (87 m)","outdoor road",{"dataset":634,"sequence":644,"environment":645},"Lawn (86 m)","outdoor lawn",[647,649,651,653],{"name":648,"methodId":5,"linkable":85,"proposed":81,"self":85},"Fovis",{"name":650,"methodId":75,"linkable":81,"proposed":81,"self":81},"DVO",{"name":652,"methodId":626,"linkable":85,"proposed":85,"self":81},"Our VO (RGB-D)",{"name":654,"methodId":626,"linkable":85,"proposed":85,"self":81},"Our VO (Lidar)",[656,658,660,662,664,666,668,669,671,673,675,677,678,680,682,684],[191,191,191,657,193,191,193,193,191],2.72,[191,191,195,659,193,191,193,193,191],5.56,[191,191,198,661,193,191,193,193,191],13.04,[191,191,201,663,193,191,193,193,191],9.97,[195,191,191,665,193,191,193,193,191],1.87,[195,191,195,667,193,191,193,193,191],8.36,[195,191,198,218,193,191,193,193,191],[195,191,201,670,193,191,193,193,191],32.07,[198,191,191,672,193,191,193,193,191],2.14,[198,191,195,674,193,191,193,193,191],1.84,[198,191,198,676,193,191,193,193,191],1.53,[198,191,201,276,193,191,193,193,191],[201,191,191,679,193,191,193,193,191],2.06,[201,191,195,681,193,191,193,193,191],1.79,[201,191,198,683,193,191,193,193,191],0.79,[201,191,201,685,193,191,193,193,191],1.73,[],[688],"Table I; Sec. VII",[],[],[692],"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",[694,700],{"group":695,"slug":696,"sourceLabel":6,"table":697,"selfRows":220,"datasets":698},"fovis2017:Fig. 4","fovis2017-fig-4","Fig. 4",[699],"authors' position-hold flight",{"group":701,"slug":702,"sourceLabel":313,"table":703,"selfRows":198,"datasets":704},"mrsmap2014:Table 2","mrsmap2014-table-2","Table 2",[322],1790510662299]