[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-85251-en":3,"doc-seo-85251-105":30,"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":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},85251,1374391974564,"Clementine","https://ap-avatar.wpscdn.com/avatar/14000253aa45c000a9e?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779874745381141002",8,"Research & Report","Distributed Agent System Fault-Tolerant Collaboration Among Embodied Agents","AI engineering is shifting from passive text generation by large language models (LLMs) to agent-driven task execution, introducing new reliability challenges for long-horizon work under limited resources and environmental uncertainty. Conventional error-elimination optimization cannot handle cumulative error propagation. This paper presents a Distributed Agent System (DAS) device-edge-cloud framework for heterogeneous embodied agents, redefining reliability as system-level fault tolerance and providing a two-layer architecture covering single-agent execution and cross-agent communication reliability. ","Distributed Agent System: Fault-Tolerant Collaboration Among Embodied Agents  \nKai Yu*, Lu Chen, Hanqi Li  \na X-LANCE Lab, School of Computer Science, Shanghai Jiao Tong University, Shanghai, 200240, China b Jiangsu Key Lab of Language Computing, Suzhou, 215028, China  \n* [Corresponding author. E-mail: kai.yu@sjtu.edu.cn](Corresponding author. E-mail: kai.yu@sjtu.edu.cn), Tel: 13917575615, Address: 800 Dongchuan Road, Minhang District, Shanghai, 200240, China  \nABSTRACT  \nAI engineering is shifting from passive text generation by large language models (LLMs) to agent-driven task execution, creating new reliability challenges for long-horizon tasks under resource constraints and environmental uncertainty. Conventional error-elimination optimization strategies fail to address cumulative error propagation. This paper proposes Distributed Agent System (DAS), a device-edge-cloud framework for fault-tolerant collaboration among heterogeneous agents. We redefine agent reliability as system-level fault tolerance rather than single-turn zero-error accuracy, and present a two-layer faulttolerance architecture: single-agent execution reliability via fault-tolerant alignment, and cross-agent communication reliability via semi-formal language protocols. This framework provides a practical engineering pathway for reliable heterogeneous embodied agents collaboration in industrial scenarios.  \nKEYWORDS  \nDistributed Agent System; Fault-Tolerant Collaboration; Embodied Agents; Reliability Alignment; SemiFormal Language Protocol; Multi-agent Collaboration  \n1. Introduction: From Generative AI to Agentic Task Execution  \nAI engineering is undergoing a fundamental shift from passive text generation dominated by large language models (LLMs) towards agent-driven, end-to-end task execution [1] . Unlike earlier application modes that rely on staged human intervention or straightforward instruction matching, next-generation agents can complete a closed loop of environmental perception, reasoning, planning, decision-making, and physical execution, thereby supporting long-horizon autonomous operations in industrial settings [2,3] . This paradigm shift also changes how AI reliability should be evaluated. Q&A-based LLM interaction is typically evaluated by single-turn accuracy, whereas autonomous industrial embodied agents must be evaluated by system-level stability and reliability across continual multi-step tasks.  \nLong-horizon autonomous execution poses new reliability challenges and has become a core bottleneck for industrial AI deployment. End-deployed lightweight embodied agents face strict constraints on computing power, storage, and bandwidth, with limited model capacity compared to cloud frontier LLMs. Dynamic disturbances in open industrial environments lead to inherent uncertainty, which cannot be fully eliminated by model scaling or data iteration. Critically, minor local errors accumulate and amplify along long task chains, potentially causing overall task failure.  \nCurrent mainstream optimization methods, including tool-parameter calibration, model confidence estimation, and multi-turn self-reflection, can mitigate model hallucinations and execution errors only in single-turn or local scenarios [4,5,6,7] . They cannot suppress the systematic amplification of errors in longhorizon tasks. Meanwhile, most multi-agent collaboration research is built on homogeneous cloud-hosted LLMs for virtual scenarios, prioritizing single-turn accuracy, and cannot adapt to heterogeneous industrial end-device deployment and physical execution [8] . To bridge this gap, we propose a distributed faulttolerant collaboration framework suitable for end-device embodied agents, replacing error elimination with system-level fault-tolerance management.  \n2. Distributed Agent System: Definition and Scope  \nA Distributed Agent System (DAS) is a distributed framework for collective intelligence, built for heterogeneous end-edge-cloud agent deployment. It balances local au","cbCaicCVYi7YHmcc","https://ap.wps.com/l/cbCaicCVYi7YHmcc","pdf",133553,3,1,5,"English","en",105,"# 1. Introduction: From Generative AI to Agentic Task Execution\n# 2. Distributed Agent System: Definition and Scope","[{\"question\":\"Why does AI reliability evaluation change for agentic embodied tasks?\",\"answer\":\"LLM Q\\u0026A is often assessed with single-turn accuracy, but embodied industrial agents require system-level stability and reliability across continual multi-step tasks.\"},{\"question\":\"What reliability problem does the paper identify in long-horizon execution?\",\"answer\":\"Minor local errors can accumulate and amplify along long task chains, potentially causing overall task failure even when individual components are improved.\"},{\"question\":\"How does DAS achieve fault-tolerant collaboration among heterogeneous agents?\",\"answer\":\"DAS redefines agent reliability as system-level fault tolerance and uses a two-layer architecture: fault-tolerant alignment for single-agent execution reliability and semi-formal language protocols for cross-agent communication reliability.\"}]",1784202087,13,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"distributed-agent-system-fault-tolerant-collaboration-among-embodied-agents","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":20},"https://docshare.wps.com/document/research-report/",{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/distributed-agent-system-fault-tolerant-collaboration-among-embodied-agents/85251/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-23","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},"Why does AI reliability evaluation change for agentic embodied tasks?","Question",{"text":75,"@type":76},"LLM Q&A is often assessed with single-turn accuracy, but embodied industrial agents require system-level stability and reliability across continual multi-step tasks.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What reliability problem does the paper identify in long-horizon execution?",{"text":80,"@type":76},"Minor local errors can accumulate and amplify along long task chains, potentially causing overall task failure even when individual components are improved.",{"name":82,"@type":73,"acceptedAnswer":83},"How does DAS achieve fault-tolerant collaboration among heterogeneous agents?",{"text":84,"@type":76},"DAS redefines agent reliability as system-level fault tolerance and uses a two-layer architecture: fault-tolerant alignment for single-agent execution reliability and semi-formal language protocols for cross-agent communication reliability.","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":25},{"code":4,"msg":5,"data":92},[93,97,101,105,109,114,119,122,127,130,134],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":22,"doc_module":4,"doc_module_name":46,"category_name":106,"show_sort_weight":107,"slug":108},"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":46,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":22,"slug":137},19,"General","general"]