[{"data":1,"prerenderedAt":98},["ShallowReactive",2],{"method-grupp2017evo":3},{"method":4,"reference":33,"equipment":47,"figures":48,"results":49},{"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":20,"limitations":23,"sensors":27,"platform":28,"estimator":29,"association":29,"timeModel":29,"deskew":29,"loopClosure":29,"globalOptimization":29,"mapRepresentation":29,"prior":29,"outputGeometry":29,"compute":29,"codeUrl":30,"codeLicense":31,"relatedVersions":32},"grupp2017evo","Grupp, 2017","evo","evo: Python package for the evaluation of odometry and SLAM.",2017,"recent","C10","evaluation_method_or_metric","evo 是評估里程計與 SLAM 軌跡的 Python 套件（GPL-3.0 以上授權），提供 evo_ape（絕對位姿誤差）與 evo_rpe（相對位姿誤差）命令列工具，並以 evo_traj 與 evo_res 做軌跡檢查、繪圖與結果比較。它支援 TUM、KITTI、EuRoC 格式與 ROS 1、ROS 2 bag，可選 SE(3) 或 Sim(3) Umeyama 對齊、僅尺度校正或原點對齊，時間關聯預設容許 0.01 s 的時間差，RPE 預設只取相鄰位姿對。作者明言它不是任何特定資料集評估協定的逐一重現；由於選項會改變數值，引用時應同時報告版本與參數（推論）。","Widely used open-source tool implementing APE\u002FRPE with configurable association, alignment and scale correction across common trajectory formats.","full_text_reviewed","software_or_dataset_record","main_body","ConSLAM 作者表示調整 evo 部分程式碼以評估其營建工地資料集上的軌跡 [trzeciak2023conslam]。HCIC 營建 RGB-D 資料集提到可用『public tools like EVO』評估，但其參考文獻 [15] 指向 Rebecq 等人 2017 的事件相機 EVO（EVO: A Geometric Approach to Event-Based 6-DOF Parallel Tracking and Mapping），而非本條 Grupp 的 evo 套件；其意圖應為本套件屬推論，引用時需註明此引文錯誤 [li2025hcic]。",[],[21,22],"Common tools for several formats and algorithmic options for association, alignment and scale adjustment (README, Why? section)","On a 20957-pose reference, evo_ape took 0.735 s versus 16.753 s for the TUM RGB-D evaluate_ate.py script with near-identical ATE RMSE (0.008119 m vs 0.008144 m) (doc\u002Fperformance.md)",[24,25,26],"README states that evo is not a 1-to-1 re-implementation of a particular evaluation protocol tailored to a specific dataset (README 'Why?')","Defaults shape the numbers: RPE uses consecutive pose pairs unless --all_pairs is set, and timestamp association uses --t_max_diff 0.01 s by default (wiki Metrics; wiki evo_traj)","(inference) Different option settings (alignment, scale, association) yield different values for the same trajectories, so results are only comparable when options are reported",[],[],"not_applicable","https:\u002F\u002Fgithub.com\u002FMichaelGrupp\u002Fevo","GPL-3.0-or-later (README and LICENSE checked)",[],{"id":5,"kind":34,"shortName":7,"title":8,"authors":35,"year":9,"venue":37,"venueType":34,"publisher":38,"volumeIssuePages":29,"doi":39,"arxivId":39,"url":30,"firstPublicDate":40,"publicationStatus":16,"metadataStatus":41,"fulltextStatus":15,"era":10,"classicReason":29,"codeUrl":30,"cluster":11,"topics":42,"mdpi":43,"verification":44,"label":6,"fulltextRoute":45,"versionRead":46,"addedByCensus":43},"software",[36],"Michael Grupp","GitHub repository (software)","self-published (GitHub \u002F PyPI)",null,"2017-09-13","metadata_partial",[11],false,"corrected","other","GitHub README at HEAD (PyPI release 1.37.1 current on 2026-09-25), wiki pages Home, Metrics, Formats, evo_traj and Plotting, and doc\u002Fperformance.md; verifier on 2026-09-25 also read the remaining wiki pages Configuration, Rerun-integration and Result-Comparison (evo_res)",[],[],{"totalRows":50,"groupCount":51,"groups":52,"others":97},2,1,[53],{"slug":54,"group":55,"sourceId":5,"sourceLabel":6,"table":56,"selfRows":50,"metrics":57,"seqs":68,"entrants":73,"cells":80,"outcomes":89,"locators":90,"hardware":92,"wordings":93,"notes":94},"grupp2017evo-text-doc-performance-md","grupp2017evo:Text doc\u002Fperformance.md","Text doc\u002Fperformance.md",[58,63],{"label":59,"unit":60,"statistic":61,"alignment":62},"APE w.r.t. translation part, rmse","m","RMSE","SE3",{"label":64,"unit":65,"statistic":66,"alignment":67},"real time of evaluation command","s","not_reported","none",[69],{"dataset":70,"sequence":71,"environment":72},"TUM RGB-D","fr2_desk","indoor",[74,77],{"name":75,"methodId":5,"linkable":76,"proposed":76,"self":76},"evo_ape --align",true,{"name":78,"methodId":79,"linkable":43,"proposed":43,"self":43},"evaluate_ate.py (TUM RGB-D tools)","sturm2012tum",[81,85,87],[82,82,82,83,84,82,84,84,82],0,0.008119,-1,[51,51,82,86,84,82,84,84,51],16.753,[82,51,82,88,84,82,84,84,51],0.735,[],[91],"doc\u002Fperformance.md",[],[],[95,96],"TUM fr2_desk ground truth (20957 poses) vs ORB trajectory; evo_ape run with --align (SE(3) Umeyama).","Wall-clock (real) time of the full evaluation command on the same inputs.",[],1790510664948]