[{"data":1,"prerenderedAt":318},["ShallowReactive",2],{"method-camvox2021":3},{"method":4,"reference":58,"equipment":82,"figures":124,"results":166},{"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":27,"sensors":32,"platform":36,"estimator":38,"association":39,"timeModel":40,"deskew":41,"loopClosure":42,"globalOptimization":43,"mapRepresentation":44,"prior":45,"outputGeometry":46,"compute":47,"codeUrl":48,"codeLicense":49,"relatedVersions":50},"camvox2021","Zhu et al., 2021","CamVox","CamVox: A Low-cost and Accurate Lidar-assisted Visual SLAM System",2021,"recent","C07","full_slam_with_global_correction","CamVox 把低成本的 Livox Horizon 固態 LiDAR 當作 ORB-SLAM2 的深度感測器：LiDAR 點先以 IMU 依各點時間校正運動畸變並轉到相機觸發時刻，再投影成與彩色影像逐像素對應的深度圖，組成 RGB-D 影格交給 ORB-SLAM2 的追蹤、局部建圖與迴圈閉合；由於 LiDAR 可量到上百公尺，深度小於 130 m 的特徵都視為近點。系統另利用 Livox 非重複掃描在靜止數秒後即可累積成高密度影像的特性，比對相機影像與 LiDAR 反射強度及深度影像的邊緣，於機器人靜止時自動做無標靶外參校正。","Uses a Livox Horizon non-repeating-scan LiDAR as the depth sensor of ORB-SLAM2 RGB-D mode (IMU-deskewed points projected to a depth image aligned with the camera, close-keypoint threshold raised to 130 m) and adds targetless automatic LiDAR-camera calibration by edge matching of accumulated reflectivity and depth images while the robot is stationary.","full_text_reviewed","peer_reviewed_published","supplementary","論文只在南方科技大學校園的戶外路線測試，以 GPS-RTK 為軌跡真值，沒有施工現場或點雲幾何精度的評估。其價值在於以低成本 Livox 固態 LiDAR 與相機組成可自動校正外參的手持或小型平台，並輸出彩色點雲，與施工現場常見的低成本掃描設備形式相近（推論）；自動無標靶校正對現場震動造成的外參偏移有參考意義。",[20],"independent_reference",[22,23,24,25,26],"On the SUSTech dataset APE RMSE 1.8 m versus 7.5 m for VINS-mono and 6.5 m for livox_horizon_loam (Table I)","Automatic calibration from a misalignment of more than 2 deg reached a cost of 6.11 versus 5.88 for the best manual calibration and worked in outdoor natural, outdoor built and underexposed indoor scenes (Sec. IV-A)","Depth-associated feature points detected reliably beyond 100 m, whereas the RealSense RGB-D camera gave no points beyond 10 m and suffered from sunlight noise (Sec. III-B)","Real-time operation on the onboard Manifold 2C computer (Sec. IV-D)","Hardware, code and dataset released; the Livox Horizon sold for 800 USD versus 10k to 80k USD for similar lidars (Sec. III-A)",[28,29,30,31],"Loop closing takes 7821.22 ms per call on the onboard computer, far longer than in ORB-SLAM2 on TUM (598.70 ms) (Table II)","Automatic calibration takes about 58 s and needs the robot to be stationary (Sec. IV-D)","Only one outdoor sequence is evaluated quantitatively; ground truth is from GPS-RTK (Sec. IV-C)","The LiDAR motion correction is described as partial in the IEEE version (Sec. III-B)",[33,34,35],"monocular rolling-shutter camera (MV-CE060-10UC)","solid-state non-repeating-scan LiDAR (Livox Horizon)","IMU (Inertial Sense uINS) used only for LiDAR motion distortion correction",[37],"wheeled UGV (AgileX Scout mini robot platform)","ORB-SLAM2 in RGB-D mode: tracking, local mapping with local bundle adjustment, loop closing and full bundle adjustment of ORB-SLAM2, fed with RGB-D frames built from the camera image and a depth image projected from IMU-corrected Livox points; keypoints with LiDAR depth below 130 m are treated as close points (Sec. III)","ORB features on the camera image with depth from the projected LiDAR depth image; extrinsic calibration by edge matching between the camera image and LiDAR reflectivity and depth images (Canny edges, edges shorter than 200 pixels and cluttered interior edges removed), a K-D-tree ICP cost with a mismatch penalty, and coordinate descent over roll, pitch and yaw in Ceres (Sec. III-C, Eq. 3 in the IEEE version)","discrete RGB-D frames at the 10 Hz hardware trigger; each LiDAR point transformed with the IMU pose at its timestamp to the LiDAR frame at the trigger time (Sec. III-B)","IMU-based correction of each LiDAR point to the trigger time of the camera image; IMU at 200 Hz synchronized with the trigger (Sec. III-A, III-B); the IEEE version describes the correction as partial","yes, ORB-SLAM2 loop closing (Sec. III; Fig. 4; Table II)","ORB-SLAM2 pose-graph optimization and full bundle adjustment after loop closure (Sec. III)","ORB-SLAM2 keyframes and map points; dense colored point cloud (RGB-D map) reconstructed from the frames for visualization (Fig. 1)","initial LiDAR-camera extrinsic refined by the automatic calibration thread when the robot is stationary; camera intrinsics assumed known (Sec. III-C)","camera trajectory and a dense RGB-colored point cloud map (Fig. 1; Fig. 6 in the IEEE numbering)","DJI Manifold 2C onboard computer with a 4-core Intel Core i7-8550U; real time; tracking 42.27 ms, mapping 252.41 ms, loop closing 7821.22 ms, IMU correction 0.89 ms and point-cloud-to-depth 16.35 ms per call; automatic calibration about 58 s, run only when stationary (Sec. IV-D; Table II)","https:\u002F\u002Fgithub.com\u002FISEE-Technology\u002FCamVox","GPL-2.0 (GitHub license metadata)",[51,55],{"relation":52,"title":53,"doi_or_url":54},"preprint","CamVox (arXiv v1, submitted version; figure numbering differs from the version of record)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2011.11357",{"relation":56,"title":57,"doi_or_url":48},"code_release","ISEE-Technology\u002FCamVox (hardware, code and dataset)",{"id":5,"kind":59,"shortName":7,"title":8,"authors":60,"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":48,"cluster":11,"topics":76,"mdpi":77,"verification":78,"label":6,"fulltextRoute":79,"versionRead":80,"addedByCensus":81},"method",[61,62,63,64,65],"Yuewen Zhu","Chunran Zheng","Chongjian Yuan","Xu Huang","Xiaoping Hong","2021 IEEE International Conference on Robotics and Automation (ICRA)","conference","IEEE","pp. 5049-5055","10.1109\u002Ficra48506.2021.9561149","2011.11357","https:\u002F\u002Fdoi.org\u002F10.1109\u002FICRA48506.2021.9561149","2020-11-23","metadata_verified","not_applicable",[11],false,"corrected","NTU institutional (Chrome)","IEEE version of record (ICRA 2021 HTML through NTU access, including Table I and II images) and arXiv v1 (2020-11-23, submitted version, CC BY 4.0)",true,[83,90,95,100,105,113,118],{"category":84,"model":85,"canonical":85,"role":86,"dataset":87,"specs":88,"locator":89},"camera","MV-CE060-10UC","method input","SUSTech dataset (CamVox)","rolling shutter camera; outputs at each 10 Hz trigger; 1520 x 568 images used","Sec. III-A; Table II",{"category":91,"model":92,"canonical":92,"role":86,"dataset":87,"specs":93,"locator":94},"lidar","Livox Horizon","non-repeating scan; detection up to 260 m under strong sunlight (cited spec); clock synced with GPS-RTK; sale price 800 USD","Sec. II; Sec. III-A",{"category":96,"model":97,"canonical":97,"role":86,"dataset":87,"specs":98,"locator":99},"imu","Inertial Sense uINS","200 Hz, synchronized with the trigger; used for LiDAR motion correction","Sec. III-A",{"category":101,"model":102,"canonical":102,"role":103,"dataset":87,"specs":104,"locator":99},"gnss","Inertial Sense uINS (GPS-RTK)","reference or ground truth","GPS-RTK recorded for ground truth",{"category":106,"model":107,"canonical":108,"role":109,"dataset":110,"specs":111,"locator":112},"rgbd","Intel Realsense D435","Intel RealSense D435","compared device",null,"mounted for comparison; no points beyond 10 m and sunlight noise outdoors","Sec. III-A; Sec. III-B",{"category":114,"model":115,"canonical":115,"role":86,"dataset":110,"specs":116,"locator":117},"platform","Agile X Scout mini","moving robot platform carrying the CamVox hardware","Sec. III-A; Fig. 3",{"category":119,"model":120,"canonical":120,"role":121,"dataset":110,"specs":122,"locator":123},"compute","DJI Manifold 2C","compute for runtime","4-core Intel Core i7-8550U; real-time CamVox","Sec. IV-D",[125,138,147,157],{"refId":5,"refLabel":6,"fig":126,"whatZh":127,"license":128,"licenseUrl":129,"sourceUrl":130,"src":131,"width":132,"height":133,"thumb":134,"thumbWidth":135,"thumbHeight":136,"modified":137},"Fig. 1 (arXiv v1)","CamVox 範例結果：機器人軌跡、稠密 RGB-D 彩色點雲地圖與局部放大","CC BY 4.0","http:\u002F\u002Fcreativecommons.org\u002Flicenses\u002Fby\u002F4.0\u002F","https:\u002F\u002Farxiv.org\u002Fhtml\u002F2011.11357v1\u002Fexample.png","\u002Ffigure-files\u002Fcamvox2021\u002Ffig-1-arxiv-v1.webp",1096,349,"\u002Ffigure-files\u002Fcamvox2021\u002Ffig-1-arxiv-v1.thumb.webp",480,153,"converted to WebP",{"refId":5,"refLabel":6,"fig":139,"whatZh":140,"license":128,"licenseUrl":129,"sourceUrl":141,"src":142,"width":143,"height":144,"thumb":145,"thumbWidth":135,"thumbHeight":146,"modified":137},"Fig. 3 (arXiv v1)","完整機器人平台與 CamVox 硬體近照（相機、Livox Horizon、IMU、GPS-RTK），以及對應的彩色影像與 LiDAR 深度影像","https:\u002F\u002Farxiv.org\u002Fhtml\u002F2011.11357v1\u002Fplatform.png","\u002Ffigure-files\u002Fcamvox2021\u002Ffig-3-arxiv-v1.webp",1162,432,"\u002Ffigure-files\u002Fcamvox2021\u002Ffig-3-arxiv-v1.thumb.webp",178,{"refId":5,"refLabel":6,"fig":148,"whatZh":149,"license":128,"licenseUrl":129,"sourceUrl":150,"src":151,"width":152,"height":153,"thumb":154,"thumbWidth":135,"thumbHeight":155,"modified":156},"Fig. 4 (arXiv v1)","CamVox SLAM 流程：ORB-SLAM2 主執行緒之外新增 RGB-D 輸入前處理執行緒與自動校正執行緒","https:\u002F\u002Farxiv.org\u002Fhtml\u002F2011.11357v1\u002FCamVox_flow.png","\u002Ffigure-files\u002Fcamvox2021\u002Ffig-4-arxiv-v1.webp",1400,459,"\u002Ffigure-files\u002Fcamvox2021\u002Ffig-4-arxiv-v1.thumb.webp",157,"resized to at most 1400 px wide and converted to WebP",{"refId":5,"refLabel":6,"fig":158,"whatZh":159,"license":128,"licenseUrl":129,"sourceUrl":160,"src":161,"width":162,"height":163,"thumb":164,"thumbWidth":135,"thumbHeight":165,"modified":137},"Fig. 7 (arXiv v1)","自動校正流程：由靜止累積的 LiDAR 反射強度與深度影像和相機灰階影像擷取邊緣後迭代匹配","https:\u002F\u002Farxiv.org\u002Fhtml\u002F2011.11357v1\u002Fcalibration_process.png","\u002Ffigure-files\u002Fcamvox2021\u002Ffig-7-arxiv-v1.webp",1345,853,"\u002Ffigure-files\u002Fcamvox2021\u002Ffig-7-arxiv-v1.thumb.webp",304,{"totalRows":167,"groupCount":168,"groups":169,"others":317},13,2,[170,263],{"slug":171,"group":172,"sourceId":5,"sourceLabel":6,"table":173,"selfRows":174,"metrics":175,"seqs":198,"entrants":202,"cells":209,"outcomes":257,"locators":258,"hardware":259,"wordings":260,"notes":261},"camvox2021-table-i","camvox2021:Table I","Table I",7,[176,181,184,187,189,192,195],{"label":177,"unit":178,"statistic":179,"alignment":180},"APE max","m","max","not_reported",{"label":182,"unit":178,"statistic":183,"alignment":180},"APE mean","mean",{"label":185,"unit":178,"statistic":186,"alignment":180},"APE median","median",{"label":188,"unit":178,"statistic":180,"alignment":180},"APE min",{"label":190,"unit":178,"statistic":191,"alignment":180},"APE rmse","RMSE",{"label":193,"unit":194,"statistic":180,"alignment":180},"APE sse","m^2 (unit not printed)",{"label":196,"unit":178,"statistic":197,"alignment":180},"APE std","std",[199],{"dataset":87,"sequence":200,"environment":201},"SUSTech route","outdoor campus roads, strong sunlight",[203,204,207],{"name":7,"methodId":5,"linkable":81,"proposed":81,"self":81},{"name":205,"methodId":206,"linkable":81,"proposed":77,"self":77},"VINS-mono","vinsmono2018",{"name":208,"methodId":110,"linkable":77,"proposed":77,"self":77},"livox_horizon_loam",[210,214,217,219,222,225,228,231,233,235,237,239,241,243,245,247,249,251,252,253,255],[211,211,211,212,213,211,213,213,211],0,3.3,-1,[211,215,211,216,213,211,213,213,211],1,1.7,[211,168,211,218,213,211,213,213,211],1.6,[211,220,211,221,213,211,213,213,211],3,0.2,[211,223,211,224,213,211,213,213,211],4,1.8,[211,226,211,227,213,211,213,213,211],5,16066.5,[211,229,211,230,213,211,213,213,211],6,0.7,[215,211,211,232,213,211,213,213,211],27.2,[215,215,211,234,213,211,213,213,211],6.7,[215,168,211,236,213,211,213,213,211],6.1,[215,220,211,238,213,211,213,213,211],2.8,[215,223,211,240,213,211,213,213,211],7.5,[215,226,211,242,213,211,213,213,211],50788.6,[215,229,211,244,213,211,213,213,211],3.5,[168,211,211,246,213,211,213,213,211],9.9,[168,215,211,248,213,211,213,213,211],6.2,[168,168,211,250,213,211,213,213,211],6.5,[168,220,211,224,213,211,213,213,211],[168,223,211,250,213,211,213,213,211],[168,226,211,254,213,211,213,213,211],101223.7,[168,229,211,256,213,211,213,213,211],1.9,[],[173],[],[],[262],"SUSTech dataset collected by the authors (outdoor campus route around the SUSTech expert apartment area) evaluated with evo against GPS-RTK (Inertial Sense uINS) ground truth; APE statistics in meters",{"slug":264,"group":265,"sourceId":5,"sourceLabel":6,"table":266,"selfRows":229,"metrics":267,"seqs":282,"entrants":295,"cells":297,"outcomes":310,"locators":311,"hardware":312,"wordings":314,"notes":315},"camvox2021-table-ii","camvox2021:Table II","Table II",[268,271,274,276,278,280],{"label":269,"unit":270,"statistic":180,"alignment":180},"Calibration time","s",{"label":272,"unit":273,"statistic":180,"alignment":180},"Tracking time","ms",{"label":275,"unit":273,"statistic":180,"alignment":180},"Mapping time",{"label":277,"unit":273,"statistic":180,"alignment":180},"Loop Closing time",{"label":279,"unit":273,"statistic":180,"alignment":180},"RGBD preprocessing: IMU correction time",{"label":281,"unit":273,"statistic":180,"alignment":180},"RGBD preprocessing: Pcd2Depth time",[283,285,287,289,291,293],{"dataset":87,"sequence":284,"environment":201},"Calibration",{"dataset":87,"sequence":286,"environment":201},"Tracking",{"dataset":87,"sequence":288,"environment":201},"Mapping",{"dataset":87,"sequence":290,"environment":201},"Loop Closing",{"dataset":87,"sequence":292,"environment":201},"RGBD preprocessing: IMU correction",{"dataset":87,"sequence":294,"environment":201},"RGBD preprocessing: Pcd2Depth",[296],{"name":7,"methodId":5,"linkable":81,"proposed":81,"self":81},[298,300,302,304,306,308],[211,211,211,299,213,211,211,213,211],58.16,[211,215,215,301,213,211,211,213,211],42.27,[211,168,168,303,213,211,211,213,211],252.41,[211,220,220,305,213,211,211,213,211],7821.22,[211,223,223,307,213,211,211,213,211],0.89,[211,226,226,309,213,211,211,213,211],16.35,[],[266],[313],"DJI Manifold 2C (4-core Intel Core i7-8550U)",[],[316],"Timing analysis: CamVox on the SUSTech dataset (1520 x 568 images, 10 Hz, 1500 ORB features) versus ORB-SLAM2 on TUM (640 x 480, 30 Hz, 1000 ORB features); per-thread time",[],1790510656523]