[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-85703-en":3,"doc-seo-85703-105":28,"detail-sidebar-cat-0-en-105":81},{"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":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":11,"language":21,"language_code":22,"site_id":23,"html_lang":22,"table_of_contents":24,"faqs":25,"seo_title":13,"seo_description":14,"update_tm":26,"read_time":27},85703,137441390410,"Hazel","https://ap-avatar.wpscdn.com/avatar/2000252f4ab5702993?_k=1776741390130283984",8,"Research & Report","OmniSCS: Omni Safety-Critical Scenario Synthesis for Autonomous Driving via a Fully Editable Driving World","Safety-critical scenario (SCS) synthesis and closed-loop simulation evaluation are essential for building robust autonomous driving systems. A core requirement is editing agent states across appearance and trajectory levels while preserving data fidelity, yet existing approaches struggle to maintain realism after scene edits and to generate high-quality SCS efficiently. OmniSCS is proposed to produce photorealistic SCS with strong physical fidelity. It includes a fully editable driving world construction stage with dual-strategy reconstruction and depth-refinement background reconstruction, and an ASCS synthesis stage supporting object insertion and trajectory editing. Experiments on nuScenes, Waymo, and KITTI show improved edited-scene fidelity and real-time closed-loop testing.","OmniSCS: Omni Safety-Critical Scenario Synthesis for Autonomous Driving via a Fully Editable Driving World  \nXiaoyun Dong 1 , Qian Xu 1 , Yang Lu 1 , Yang Lou 1 , Yung-Hui Li2 , and Jianping Wang 1 1  \n1 City University of Hong Kong  \n2Hon Hai Research Institute  \narXiv :2607 .09764v 1 [ cs .RO] 7 Jul 2026  \n(a) Driving scene editing with agent modification and removal  \n(b) Appearance-level safety-critical scenario synthesis (c) Behavior-level safety-critical scenario synthesis  \nFig. 1: OmniSCS enables a fully editable driving world, supporting high-fidelity appearance-level and behavior-level SCS synthesis for closed-loop testing. Please visit our project page ([https://omniscs.github.io/](https://omniscs.github.io/)) for more results.  \nAbstract—The synthesis of safety-critical scenarios (SCS) and their evaluation through closed-loop simulations are crucial for developing robust autonomous driving systems. A key aspect of this process involves editing agent states in both appearance and trajectory levels within existing scenes. However, current methods struggle to preserve data fidelity after scene editing and fail to efficiently generate high-quality SCS through such modifications. To overcome these limitations, we propose OmniSCS, an innovative system that generates photorealistic SCS with high physical fidelity while enabling closed-loop testing in synthetic environments. OmniSCS comprises two key modules: 1) A Fully Editable Driving World Construction module that maintains high-fidelity agent appearance and background during scene editing via dual-strategy agent reconstruction and depth-refinement background reconstruction methods. 2) ASCS Synthesis module that facilitates object insertion and agent trajectory editing to synthesize diverse SCS while preserving data fidelity. Experiments on nuScenes, Waymo, and KITTI datasets show that OmniSCS outperforms state-of-the-art methods in edited scene fidelity. We further validate its ability to enhance autonomous driving algorithms and support real-time (13Hz) closed-loop testing. Overall, OmniSCS provides a safer, more effective, and cost-efficient solution for SCS optimization and testing in autonomous driving.  \nI. INTRODUCTION  \nAlthough recent advancements in modular and end-to-end (E2E) frameworks have enabled autonomous vehicles (AVs) to perform reliably in most routine driving scenarios [1], [2], they still struggle in safety-critical scenarios (SCS) . These high-risk situations—which represent only 1% of driving data [3]—are difficult to address due to data scarcity and the inherent dangers of real-world closed-loop testing. As a  \nresult, high-fidelity data synthesis of safety-critical scenarios for effective closed-loop testing methods remains a crucial and yet underexplored area of research.  \nFollowing the classification in this field [4], [5], SCS are categorized into two levels: 1) Appearance-level, i.e., rare and unannotated objects, like excavators, forklifts, etc., common objects appearing in an unusual way or locations, like a truck overturned on the ground, resulting in failures in both perception module and E2E algorithms; 2) Behavior-level, i.e., trajectory collision risk between the ego vehicle and surrounding vehicles, resulting in motion planning failuresin both planning module and E2E algorithms.  \nRecent research has explored reconstructing 3D scenes with high visual and geometric fidelity from recorded driving videos using Neural Radiance Fields (NeRFs) [6] and Gaussian Splatting (GS) [7], [8], [9], [10] . However, these approaches primarily reproduce data patterns from driving logs with similar camera perspectives, limiting generalization to unseen viewpoints. When agent trajectories deviate significantly from driving logs (Fig. 1(a)), artifacts such as blurring and ghosting degrade visual quality and hinder subsequent closed-loop evaluation. Although generative model-based approaches [11], [12], [13], [14] are capable of synthesizing diverse dri","cbCaiu4Zf9ZRuUys","https://ap.wps.com/l/cbCaiu4Zf9ZRuUys","pdf",2722412,1,"English","en",105,"# Introduction\n## Safety-critical scenario levels\n## Scene reconstruction limitations\n## Key challenges\n## Proposed approach: OmniSCS","[{\"question\":\"How does OmniSCS generate safety-critical scenarios after editing?\",\"answer\":\"The ASCS Synthesis module supports object insertion and agent trajectory editing to synthesize diverse appearance-level and behavior-level SCS while preserving data fidelity for reliable closed-loop testing.\"}]",1784205700,20,{"code":4,"msg":29,"data":30},"ok",{"site_id":23,"language":22,"slug":31,"title":13,"keywords":32,"description":14,"schema_data":33,"social_meta":76,"head_meta":78,"extra_data":80,"updated_unix":26},"omniscs-omni-safety-critical-scenario-synthesis-for-autonomous-driving-via-a-fully-editable-driving-world","",{"@graph":34,"@context":75},[35,52,66],{"@type":36,"itemListElement":37},"BreadcrumbList",[38,42,46,49],{"item":39,"name":40,"@type":41,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":43,"name":44,"@type":41,"position":45},"https://docshare.wps.com/document/","Document",2,{"item":47,"name":12,"@type":41,"position":48},"https://docshare.wps.com/document/research-report/",3,{"item":50,"name":13,"@type":41,"position":51},"https://docshare.wps.com/document/omniscs-omni-safety-critical-scenario-synthesis-for-autonomous-driving-via-a-fully-editable-driving-world/85703/",4,{"url":50,"name":13,"@type":53,"author":54,"headline":13,"publisher":56,"fileFormat":59,"inLanguage":22,"description":14,"dateModified":60,"datePublished":60,"encodingFormat":59,"isAccessibleForFree":61,"interactionStatistic":62},"DigitalDocument",{"name":9,"@type":55},"Person",{"url":39,"name":57,"@type":58},"DocShare","Organization","application/pdf","2026-07-16",true,{"@type":63,"interactionType":64,"userInteractionCount":4},"InteractionCounter",{"@type":65},"ViewAction",{"@type":67,"mainEntity":68},"FAQPage",[69],{"name":70,"@type":71,"acceptedAnswer":72},"How does OmniSCS generate safety-critical scenarios after editing?","Question",{"text":73,"@type":74},"The ASCS Synthesis module supports object insertion and agent trajectory editing to synthesize diverse appearance-level and behavior-level SCS while preserving data fidelity for reliable closed-loop testing.","Answer","https://schema.org",{"og:url":50,"og:type":77,"og:title":13,"og:site_name":57,"og:description":14},"article",{"robots":79,"canonical":50},"index,follow",{"doc_id":7,"site_id":23},{"code":4,"msg":5,"data":82},[83,87,91,95,100,105,110,113,117,120,124],{"id":20,"doc_module":4,"doc_module_name":44,"category_name":84,"show_sort_weight":85,"slug":86},"Story & Novel",90,"story-novel",{"id":45,"doc_module":4,"doc_module_name":44,"category_name":88,"show_sort_weight":89,"slug":90},"Literature",80,"literature",{"id":51,"doc_module":4,"doc_module_name":44,"category_name":92,"show_sort_weight":93,"slug":94},"Exam",70,"exam",{"id":96,"doc_module":4,"doc_module_name":44,"category_name":97,"show_sort_weight":98,"slug":99},5,"Comic",60,"comic",{"id":101,"doc_module":4,"doc_module_name":44,"category_name":102,"show_sort_weight":103,"slug":104},6,"Technology",50,"technology",{"id":106,"doc_module":4,"doc_module_name":44,"category_name":107,"show_sort_weight":108,"slug":109},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":44,"category_name":12,"show_sort_weight":111,"slug":112},30,"research-report",{"id":114,"doc_module":4,"doc_module_name":44,"category_name":115,"show_sort_weight":27,"slug":116},9,"Religion & Spirituality","religion-spirituality",{"id":27,"doc_module":4,"doc_module_name":44,"category_name":118,"show_sort_weight":27,"slug":119},"World Cup","world-cup",{"id":121,"doc_module":4,"doc_module_name":44,"category_name":122,"show_sort_weight":121,"slug":123},10,"Lifestyle","lifestyle",{"id":125,"doc_module":4,"doc_module_name":44,"category_name":126,"show_sort_weight":96,"slug":127},19,"General","general"]