[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-83741-en":3,"doc-seo-83741-105":29,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},83741,549758252649,"Ivy","https://ap-avatar.wpscdn.com/avatar/8000253669c5317157?_k=1778319167496531819",6,"Technology","EvoEye Self-Evolving Runtime Monitoring for Autonomous Driving Systems","Runtime monitoring detects impending hazards in autonomous driving systems, yet existing monitors are limited by either fixed rule coverage or dependence on initial training data, which can leave substantial prediction errors. EvoEye identifies the current monitor’s errors, generates informative executions, and updates the monitor through self-evolution. FusionMonitor learns cross-module temporal interactions for collision prediction, while BlindSpotEvolver turns prediction errors into search guidance and uses density-aware mutation to acquire valuable executions. Experiments on Baidu Apollo with CARLA show strong recall and faster warning times.","EvoEye: Self-Evolving Runtime Monitoring for Autonomous Driving Systems  \nMingfei Cheng, Lionel Briand, Fellow, IEEE, and Xiaofei Xie  \narXiv :2607 .03755v2 [ cs . SE] 7 Jul 2026  \nAbstract—Runtime monitoring is essential for detecting impending hazards in autonomous driving systems (ADSs). However, existing ADS runtime monitors have fixed detection capabilities: rule-based monitors cover only manually specified hazards, while learning-based monitors depend heavily on their initial training data and may retain substantial prediction errors. We therefore propose EvoEye, which identifies the current monitor’s errors, generates informative executions accordingly, and updates the monitor through self-evolution. To enable effective self-evolution, EvoEye combines a capable runtime monitor with targeted scenario acquisition. FusionMonitor learns cross-module temporal interactions for collision prediction, while BlindSpotEvolver converts current prediction errors into search guidance and uses density-aware mutation to acquire informative executions for subsequent monitor updates.  \nWe evaluate EvoEye on Baidu Apollo with CARLA in representative highway and urban scenarios. FusionMonitor improves frame-level Recall by up to 37.8 percentage points at a false positive rate of 0.05, with 2.49 ms latency and 2.8–4.2 seconds of median warning time. Under the same budget, BlindSpotEvolveroutperforms uniform and violation-oriented sampling by up to  \n13.2 F1 points on previously missed unsafe contexts.  \nI. INTRODUCTION  \nAutonomous driving systems (ADSs) have advanced rapidly in recent years and offer the potential to improve transportation safety, accessibility, and efficiency [1] . An ADS comprises the hardware and software that collectively perform the dynamic driving task on a sustained basis. In a typical modular architecture, observations from cameras, LiDAR, radar, and positioning sensors are processed to perceive and understand the surrounding environment; the resulting information is then used for behavior prediction, decision-making, motion planning, and vehicle control [2] . Despite substantial progress, assuring the safety of ADSs remains challenging because they operate in open and highly dynamic environments and rely on complex, partly learning-enabled components whose behaviors cannot be exhaustively validated before deployment [3], [4] . Runtime monitoring therefore serves as an important complementary safeguard by continuously assessing ADS executions and detecting emerging hazards that escape design-time assurance [5] . Existing runtime monitoring approaches for ADSs broadly rely on either rule-based safety indicators or learning-based misbehavior predictors. Rule-based approaches use predefined indicators, such as Time-to-Collision (TTC) [6] and  \nMingfei Cheng and Xiaofei Xie are with the School of Computing and Information Systems, Singapore Management University, Singapore.(E-mail: [mfcheng.2022@smu.edu.sg](mfcheng.2022@smu.edu.sg), [xfxie@smu.edu.sg](xfxie@smu.edu.sg)).  \nLionel Briand is with the University of Ottawa, Canada, and Lero Research Ireland Centre for Software Research, University of Limerick, Ireland. (E-mail: [lbriand@uottawa.ca](lbriand@uottawa.ca))  \nResponsibility-Sensitive Safety (RSS) [7], to determine whether current vehicle interactions violate specific safety conditions. Although efficient and interpretable, these indicators are typically designed for particular conflict patterns and rely on a limited set of kinematic relationships, making it difficult to characterize compound traffic interactions and failures emerging across the full ADS stack. Learning-based monitors [8], [9] provide greater expressive power by learning complex temporal patterns directly from execution data, rather than relying on manually specified safety rules. This enables them to integrate high-dimensional runtime signals and identify risk patterns that may not be captured by individual safety indicators. However, existing ","cbCaibT70WqgSfyu","https://ap.wps.com/l/cbCaibT70WqgSfyu","pdf",1396923,1,12,"English","en",105,"# Introduction\n## Background on runtime monitoring for ADS\n## Challenges with fixed rule-based and learning-based monitors\n## Motivation for self-evolving runtime monitoring\n## Contributions of EvoEye","[{\"question\":\"What problem does EvoEye address in autonomous driving runtime monitoring?\",\"answer\":\"Existing ADS runtime monitors have fixed detection capabilities, either limited by manually specified rules or constrained by errors and blind spots from initial training data. EvoEye aims to improve monitoring coverage by actively collecting informative scenarios and updating the monitor.\"},{\"question\":\"How does FusionMonitor improve collision-related warning performance?\",\"answer\":\"FusionMonitor learns cross-module temporal interactions to better model collision prediction at the system level, improving recall while controlling false positives.\"},{\"question\":\"How does BlindSpotEvolver decide which executions to acquire for updating the monitor?\",\"answer\":\"BlindSpotEvolver converts current prediction errors into search guidance, then applies density-aware mutation to generate informative executions, outperforming uniform and violation-oriented sampling under the same budget.\"}]",1784190147,30,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"evoeye-self-evolving-runtime-monitoring-for-autonomous-driving-systems","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/technology/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/evoeye-self-evolving-runtime-monitoring-for-autonomous-driving-systems/83741/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does EvoEye address in autonomous driving runtime monitoring?","Question",{"text":75,"@type":76},"Existing ADS runtime monitors have fixed detection capabilities, either limited by manually specified rules or constrained by errors and blind spots from initial training data. EvoEye aims to improve monitoring coverage by actively collecting informative scenarios and updating the monitor.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does FusionMonitor improve collision-related warning performance?",{"text":80,"@type":76},"FusionMonitor learns cross-module temporal interactions to better model collision prediction at the system level, improving recall while controlling false positives.",{"name":82,"@type":73,"acceptedAnswer":83},"How does BlindSpotEvolver decide which executions to acquire for updating the monitor?",{"text":84,"@type":76},"BlindSpotEvolver converts current prediction errors into search guidance, then applies density-aware mutation to generate informative executions, outperforming uniform and violation-oriented sampling under the same budget.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,113,118,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":111,"slug":112},50,"technology",{"id":114,"doc_module":4,"doc_module_name":45,"category_name":115,"show_sort_weight":116,"slug":117},7,"Healthcare",40,"healthcare",{"id":119,"doc_module":4,"doc_module_name":45,"category_name":120,"show_sort_weight":28,"slug":121},8,"Research & Report","research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]