[{"data":1,"prerenderedAt":365},["ShallowReactive",2],{"method-supereight2018":3},{"method":4,"reference":57,"equipment":82,"figures":100,"results":101},{"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":34,"platform":36,"estimator":39,"association":40,"timeModel":41,"deskew":42,"loopClosure":43,"globalOptimization":43,"mapRepresentation":44,"prior":43,"outputGeometry":45,"compute":46,"codeUrl":47,"codeLicense":48,"relatedVersions":49},"supereight2018","Vespa et al., 2018","supereight","Efficient Octree-Based Volumetric SLAM Supporting Signed-Distance and Occupancy Mapping",2018,"recent","C08","odometry_with_local_mapping","supereight 提出以八元樹（octree）為空間索引的稠密體積 SLAM 框架。最底層以 8×8×8 體素區塊為單位，並以 Morton 編碼排序與逐層遮罩做免鎖的平行配置，再預先計算三線性內插的查詢順序，使八元樹在 CPU 上的融合與射線投射效率接近 InfiniTAM 的雜湊表。同一資料結構可存放 TSDF，也可存放機率佔據（occupancy）地圖；佔據地圖改寫 Loop 等人的 b-spline 雜訊模型，改用對數勝算累加、機率截斷與依時間遺忘的更新，使其適用於增量 SLAM，並明確表示已觀測的空區。追蹤採 KinectFusion 式只用深度的點對面 ICP，作者並示範佔據地圖可直接供 Informed RRT* 路徑規劃查詢。","Octree-based dense volumetric SLAM on the CPU: Morton-coded octree with 8^3 voxel-block leaves and lock-free parallel allocation that stores either a TSDF or a SLAM-ready probabilistic occupancy field (b-spline noise model, log-odds, clamping, time-windowed forgetting), with depth-only KinectFusion ICP tracking and direct use of the same map for RRT* planning.","full_text_reviewed","peer_reviewed_published","background","論文未在施工現場測試，評估為 TUM RGB-D 與 ICL-NUIM 室內序列。同一個八元樹地圖同時支援追蹤與路徑規劃，且明確區分已觀測空區與未知區域，對室內工地巡檢機器人的規劃有參考價值（推論）。corpus 中 vizzo2022vdbfusion 比較的 SuperEight 結果數值，依該紀錄引自 Wang 等人的 LiDAR 擴充版本，並非本文原始的 RGB-D 版本。",[20,21],"public_benchmark","simulation",[23,24,25,26,27],"Tracking accuracy on par with InfiniTAM on ICL-NUIM and TUM with 1 cm voxels and depth-only tracking; the occupancy pipeline was best on ICL LR 3 and TUM fr3 office (Table I)","Octree TSDF fusion and ray-casting time comparable to or faster than InfiniTAM voxel hashing (Sec. V-B, Fig. 6)","Memory 1.95 to 13.77% (TSDF) and 3.01 to 22.52% (occupancy) of a pre-allocated grid at the same resolution (Table II)","Planning queries faster than OctoMap: 12.6 ms versus 17.7 ms mean to the first feasible RRT* path and 1.57 ms versus 2.06 ms for trajectory optimization (Tables III and IV)","Runs at 10 to 40 Hz on a quad-core CPU without GPU (Abstract)",[29,30,31,32,33],"All compared systems lost track on TUM fr1 floor and fr1 plant; the TSDF pipeline reached 0.758 m ATE on ICL LR 3 (Table I)","Occupancy mapping is more expensive than TSDF fusion because of b-spline sampling, log-odds updates and explicit free-space storage (Sec. V-B)","Depth-only ICP tracking without loop closure; combined geometric and photometric tracking left to future work (Sec. V-A)","Runtime comparisons are between different code bases, so part of the differences come from implementation (Sec. V-B)","Planning is shown as query timings on the map, not in flight; drone integration is future work (Sec. V-C, VI)",[35],"RGB-D camera, depth only (TUM RGB-D real sequences and ICL-NUIM synthetic sequences; sensor models not named)",[37,38],"public RGB-D sequences (TUM RGB-D real sequences; capture platform not described in the paper)","simulation (ICL-NUIM living room; offline MAV path-planning queries on the map)","KinectFusion-style frame-to-model point-to-plane ICP solved by Gauss-Newton on depth only, against vertex and normal maps ray-cast from the octree map","Projective data association between the current depth frame and the ray-cast model prediction","discrete poses","not_applicable (depth camera)","none","Morton-coded octree whose leaves are 8 x 8 x 8 voxel blocks allocated in parallel from a memory pool; the field type is generic: TSDF, or log-odds occupancy using a quadratic b-spline depth-noise model (sigma proportional to range squared, 4 cm at 2 m), clamping to [0.03, 0.97] and a time-windowed forgetting update (tau = 5 s)","TSDF or occupancy octree map at 1 cm finest resolution (surfaces as zero crossings), usable directly for sampling-based path planning","CPU only; experiments in SLAMBench on a Skylake i7-6700HQ with 16 GB; 10 to 40 Hz on a modern quad-core CPU (Abstract); planning timings on an Intel Core i7-6600U at 2.60 GHz","https:\u002F\u002Fgithub.com\u002Femanuelev\u002Fsupereight","BSD-3-Clause for the core library with some MIT-licensed files (stated in README; no LICENSE file)",[50,54],{"relation":51,"title":52,"doi_or_url":53},"accepted_manuscript","Efficient octree-based volumetric SLAM supporting signed-distance and occupancy mapping (Imperial Spiral, EVespaRAL_final.pdf)","http:\u002F\u002Fhdl.handle.net\u002F10044\u002F1\u002F55715",{"relation":55,"title":56,"doi_or_url":47},"code_release","emanuelev\u002Fsupereight",{"id":5,"kind":58,"shortName":7,"title":8,"authors":59,"year":9,"venue":66,"venueType":67,"publisher":68,"volumeIssuePages":69,"doi":70,"arxivId":71,"url":72,"firstPublicDate":73,"publicationStatus":16,"metadataStatus":74,"fulltextStatus":15,"era":10,"classicReason":75,"codeUrl":47,"cluster":11,"topics":76,"mdpi":77,"verification":78,"label":6,"fulltextRoute":79,"versionRead":80,"addedByCensus":81},"method",[60,61,62,63,64,65],"Emanuele Vespa","Nikolay Nikolov","Marius Grimm","Luigi Nardi","Paul H. J. Kelly","Stefan Leutenegger","IEEE Robotics and Automation Letters","journal","IEEE","3(2):1144-1151","10.1109\u002Flra.2018.2792537",null,"https:\u002F\u002Fapi.crossref.org\u002Fworks\u002F10.1109\u002FLRA.2018.2792537","2018-01-12","metadata_verified","not_applicable",[11],false,"corrected","NTU institutional (Chrome)","IEEE Xplore HTML full text of the version of record (RA-L 3(2), April 2018) with Tables I to IV viewed as publisher images; references and table text cross-checked against the accepted manuscript text on Imperial Spiral",true,[83,89,93],{"category":84,"model":85,"canonical":85,"role":86,"dataset":71,"specs":87,"locator":88},"compute","Skylake i7-6700HQ CPU with 16GB of memory","compute for runtime","Ubuntu 16.10, frequency scaling disabled, GCC 5.4.1; SLAMBench framework","Sec. V",{"category":84,"model":90,"canonical":90,"role":86,"dataset":71,"specs":91,"locator":92},"Intel Core i7-6600U CPU at 2.60 GHz","used for the path-planning timings (GCC 5.4.0)","Sec. V-C",{"category":94,"model":95,"canonical":95,"role":96,"dataset":97,"specs":98,"locator":99},"other","high frequency motion capture system (model not reported)","reference or ground truth","TUM RGB-D","trajectory ground truth of the TUM RGB-D sequences","Sec. V-A",[],{"totalRows":102,"groupCount":103,"groups":104,"others":364},42,4,[105,219,288,332],{"slug":106,"group":107,"sourceId":5,"sourceLabel":6,"table":108,"selfRows":109,"metrics":110,"seqs":116,"entrants":139,"cells":147,"outcomes":212,"locators":214,"hardware":215,"wordings":216,"notes":217},"supereight2018-table-i","supereight2018:Table I","Table I",20,[111],{"label":112,"unit":113,"statistic":114,"alignment":115},"ATE (m)","m","RMSE","not_reported",[117,121,123,125,127,129,131,133,135,137],{"dataset":118,"sequence":119,"environment":120},"ICL-NUIM","ICL_LR_0","synthetic living room (ICL) and real indoor office scenes (TUM)",{"dataset":118,"sequence":122,"environment":120},"ICL_LR_1",{"dataset":118,"sequence":124,"environment":120},"ICL_LR_2",{"dataset":118,"sequence":126,"environment":120},"ICL_LR_3",{"dataset":97,"sequence":128,"environment":120},"TUM_fr1_xyz",{"dataset":97,"sequence":130,"environment":120},"TUM_fr1_floor",{"dataset":97,"sequence":132,"environment":120},"TUM_fr1_plant",{"dataset":97,"sequence":134,"environment":120},"TUM_fr1_desk",{"dataset":97,"sequence":136,"environment":120},"TUM_fr2_desk",{"dataset":97,"sequence":138,"environment":120},"TUM_fr3_office",[140,142,144],{"name":141,"methodId":5,"linkable":81,"proposed":81,"self":81},"TSDF (octree TSDF fusion, ours)",{"name":143,"methodId":5,"linkable":81,"proposed":81,"self":81},"OFusion (octree occupancy fusion, ours)",{"name":145,"methodId":146,"linkable":81,"proposed":77,"self":77},"InfiniTAM [16] (default depth-only tracker)","infinitam2015",[148,152,155,158,160,162,164,166,168,170,173,175,177,179,181,183,185,186,187,189,190,191,194,196,198,201,203,205,208,210],[149,149,149,150,151,149,151,151,149],0,0.0113,-1,[153,149,149,154,151,149,151,151,149],1,0.0305,[156,149,149,157,151,149,151,151,149],2,0.3052,[149,149,153,159,151,149,151,151,149],0.0117,[153,149,153,161,151,149,151,151,149],0.0207,[156,149,153,163,151,149,151,151,149],0.0214,[149,149,156,165,151,149,151,151,149],0.004,[153,149,156,167,151,149,151,151,149],0.005,[156,149,156,169,151,149,151,151,149],0.1725,[149,149,171,172,151,149,151,151,149],3,0.7582,[153,149,171,174,151,149,151,151,149],0.0786,[156,149,171,176,151,149,151,151,149],0.4858,[149,149,103,178,151,149,151,151,149],0.0295,[153,149,103,180,151,149,151,151,149],0.0293,[156,149,103,182,151,149,151,151,149],0.0273,[149,149,184,71,149,149,151,151,149],5,[153,149,184,71,149,149,151,151,149],[156,149,184,71,149,149,151,151,149],[149,149,188,71,149,149,151,151,149],6,[153,149,188,71,149,149,151,151,149],[156,149,188,71,149,149,151,151,149],[149,149,192,193,151,149,151,151,149],7,0.103,[153,149,192,195,151,149,151,151,149],0.0995,[156,149,192,197,151,149,151,151,149],0.0647,[149,149,199,200,151,149,151,151,149],8,0.0641,[153,149,199,202,151,149,151,151,149],0.0902,[156,149,199,204,151,149,151,151,149],0.0598,[149,149,206,207,151,149,151,151,149],9,0.0686,[153,149,206,209,151,149,151,151,149],0.0604,[156,149,206,211,151,149,151,151,149],0.0996,[213],"tracking failure",[108],[],[],[218],"ATE RMSE (Euclidean distance between ground-truth and estimated positions) on ICL-NUIM living room and TUM RGB-D; 1 cm finest voxels, depth-only tracking, same parameters throughout; 'x' = tracking failure",{"slug":220,"group":221,"sourceId":5,"sourceLabel":6,"table":222,"selfRows":223,"metrics":224,"seqs":228,"entrants":246,"cells":249,"outcomes":282,"locators":283,"hardware":284,"wordings":285,"notes":286},"supereight2018-table-ii","supereight2018:Table II","Table II",16,[225],{"label":226,"unit":227,"statistic":115,"alignment":43},"relative memory consumption compared to a pre-allocated grid","%",[229,232,234,236,238,240,242,244],{"dataset":118,"sequence":230,"environment":231},"LR_0","indoor",{"dataset":118,"sequence":233,"environment":231},"LR_1",{"dataset":118,"sequence":235,"environment":231},"LR_2",{"dataset":118,"sequence":237,"environment":231},"LR_3",{"dataset":97,"sequence":239,"environment":231},"fr1_xyz",{"dataset":97,"sequence":241,"environment":231},"fr1_desk",{"dataset":97,"sequence":243,"environment":231},"fr2_desk",{"dataset":97,"sequence":245,"environment":231},"fr3_office",[247,248],{"name":141,"methodId":5,"linkable":81,"proposed":81,"self":81},{"name":143,"methodId":5,"linkable":81,"proposed":81,"self":81},[250,252,254,256,258,260,262,264,266,268,270,272,274,276,278,280],[149,149,149,251,151,149,151,151,149],7.67,[153,149,149,253,151,149,151,151,149],11.15,[149,149,153,255,151,149,151,151,149],8.45,[153,149,153,257,151,149,151,151,149],13.68,[149,149,156,259,151,149,151,151,149],13.77,[153,149,156,261,151,149,151,151,149],22.52,[149,149,171,263,151,149,151,151,149],13.33,[153,149,171,265,151,149,151,151,149],17.68,[149,149,103,267,151,149,151,151,149],1.95,[153,149,103,269,151,149,151,151,149],3.01,[149,149,184,271,151,149,151,151,149],7.7,[153,149,184,273,151,149,151,151,149],8.81,[149,149,188,275,151,149,151,151,149],10.15,[153,149,188,277,151,149,151,151,149],17.7,[149,149,192,279,151,149,151,151,149],12.5,[153,149,192,281,151,149,151,151,149],17.95,[],[222],[],[],[287],"Memory of the octree maps relative to a statically pre-allocated grid covering the same area at the same resolution (as in KinectFusion)",{"slug":289,"group":290,"sourceId":5,"sourceLabel":6,"table":291,"selfRows":171,"metrics":292,"seqs":303,"entrants":307,"cells":313,"outcomes":325,"locators":326,"hardware":327,"wordings":329,"notes":330},"supereight2018-table-iii","supereight2018:Table III","Table III",[293,297,300],{"label":294,"unit":295,"statistic":296,"alignment":43},"planning time (time)","ms","mean",{"label":298,"unit":295,"statistic":299,"alignment":43},"planning time (std dev)","std",{"label":301,"unit":295,"statistic":302,"alignment":43},"planning time (max)","max",[304],{"dataset":305,"sequence":115,"environment":306},"not_reported (map built by the SLAM system)","indoor map, simulated MAV planning",[308,310],{"name":309,"methodId":5,"linkable":81,"proposed":81,"self":81},"OFusion (octree occupancy map, ours)",{"name":311,"methodId":312,"linkable":81,"proposed":77,"self":77},"Octomap","hornung2013octomap",[314,316,318,320,321,323],[149,149,149,315,151,149,149,151,149],12.6,[149,153,149,317,151,149,149,151,149],14.6,[149,156,149,319,151,149,149,151,149],109.6,[153,149,149,277,151,149,149,151,149],[153,153,149,322,151,149,149,151,149],11.4,[153,156,149,324,151,149,149,151,149],113.2,[],[291],[328],"Intel Core i7-6600U at 2.60 GHz",[],[331],"Time to find the first feasible straight-line path with Informed RRT* (OMPL) for an obstructed 2.83 m start-goal distance, 1 cm map, averaged over 10,000 executions",{"slug":333,"group":334,"sourceId":5,"sourceLabel":6,"table":335,"selfRows":171,"metrics":336,"seqs":340,"entrants":342,"cells":345,"outcomes":358,"locators":359,"hardware":360,"wordings":361,"notes":362},"supereight2018-table-iv","supereight2018:Table IV","Table IV",[337,338,339],{"label":294,"unit":295,"statistic":296,"alignment":43},{"label":298,"unit":295,"statistic":299,"alignment":43},{"label":301,"unit":295,"statistic":302,"alignment":43},[341],{"dataset":305,"sequence":115,"environment":306},[343,344],{"name":309,"methodId":5,"linkable":81,"proposed":81,"self":81},{"name":311,"methodId":312,"linkable":81,"proposed":77,"self":77},[346,348,350,352,354,356],[149,149,149,347,151,149,149,151,149],1.57,[149,153,149,349,151,149,149,151,149],0.59,[149,156,149,351,151,149,149,151,149],3.47,[153,149,149,353,151,149,149,151,149],2.06,[153,153,149,355,151,149,149,151,149],0.78,[153,156,149,357,151,149,149,151,149],4.17,[],[335],[328],[],[363],"Time for linear optimization of a collision-free polynomial trajectory from the RRT* plan, averaged over 1,000 executions",[],1790510664444]