[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-327705-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-327705-en":130},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","embodied-operators-and-benchmarking-toward-reusable-and-deployable-embodied-intelligence-systems","Embodied Operators and Benchmarking - Toward Reusable and Deployable Embodied Intelligence Systems","","Embodied intelligence systems require more than end-to-end policy networks; they depend on reusable modules that convert multimodal observations, robot states, demonstrations, and task context into structured representations, decisions, trajectories, control references, and services. The work defines these modules as embodied operators with clear task semantics, standardized input-output contracts, deployability, reusability, and multi-layer optimizability, then builds a taxonomy of five operator categories and a multi-dimensional benchmarking framework emphasizing correctness, efficiency, stability, portability, compatibility, reliability, and downstream utility.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/embodied-operators-and-benchmarking-toward-reusable-and-deployable-embodied-intelligence-systems/327705/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/embodied-operators-and-benchmarking-toward-reusable-and-deployable-embodied-intelligence-systems/327705.png","ImageObject",300,407,{"name":42,"@type":43},"Liam","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-26","2026-09-21",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",8,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What problem do embodied operators address in embodied intelligence pipelines?","Question",{"text":63,"@type":64},"They provide reusable functional modules with explicit task semantics and standardized input-output interfaces that transform raw multimodal signals and intermediate system states into structured representations, decisions, and executable outputs.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How are embodied operators categorized in this work?",{"text":68,"@type":64},"They are organized into five major categories: detection and segmentation, spatial localization and 3D understanding, hand motion recovery, embodied foundation models and task-decision, and planning/control/system support operators.",{"name":70,"@type":61,"acceptedAnswer":71},"How does the proposed benchmark evaluate embodied operators compared with conventional evaluation?",{"text":72,"@type":64},"It extends beyond single-model correctness to consider end-to-end efficiency, resource usage, temporal stability, portability, interface compatibility, deployment reliability, and downstream task utility.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},327705,1790507211,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,114,119,122,126],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":55,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":112,"slug":113},30,"research-report",{"id":115,"doc_module":4,"doc_module_name":25,"category_name":116,"show_sort_weight":117,"slug":118},9,"Religion & Spirituality",20,"religion-spirituality",{"id":117,"doc_module":4,"doc_module_name":25,"category_name":120,"show_sort_weight":117,"slug":121},"World Cup","world-cup",{"id":123,"doc_module":4,"doc_module_name":25,"category_name":124,"show_sort_weight":123,"slug":125},10,"Lifestyle","lifestyle",{"id":127,"doc_module":4,"doc_module_name":25,"category_name":128,"show_sort_weight":97,"slug":129},19,"General","general",{"code":4,"msg":82,"data":131},{"doc_id":79,"user_id":132,"nickname":42,"user_avatar":133,"doc_module":4,"category_id":55,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":112,"language":139,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":140,"faqs":141,"seo_title":142,"seo_description":12,"update_tm":143,"read_time":144,"preview_page_count":4},8796095461564,"https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d","arXiv :2607 .03283v 1 [ cs .AI] 3 Jul 2026  \nEmbodied Operators and Benchmarking: Toward Reusable and Deployable Embodied Intelligence Systems  \n1AI Infra Team at JDT 2 Tsinghua University 3 Tianjin University 3 Beihang University  \nFull author list in Contributions  \nAbstract  \nEmbodied intelligence systems require not only end-to-end policy models, but also a large number of reusable functional modules that transform multimodal observations, robot states, human demonstrations, and task contexts into structured representations, decisions, trajectories, control references, and system services. This work defines such modules as embodied operators and studies them as independent yet composable units within embodied intelligence pipelines. We first clarify the definition boundary of embodied operators, emphasizing their task semantics, standardized input-output contracts, deployability, reusability, and multi-layer optimizability. We then construct a taxonomy covering five major categories: detection and segmentation operators, spatial localization and 3D understanding operators, hand motion recovery operators, embodied foundation models and task-decision operators, and planning, control, and system support operators. For each category, we summarize representative functions, technical paradigms, application roles, and practical limitations in manipulation-oriented embodied systems. Beyond operator taxonomy, we further propose a multi-dimensional benchmark framework for embodied operators. Unlike conventional single-model evaluation, the proposed benchmark considers correctness, end-to-endefficiency, resource usage, temporal stability, portability, interface compatibility, deployment reliability, and downstream task utility. We also discuss operator acceleration from a workflow-level perspective and analyze open challenges in operator composition, data standardization, world models, VLA safety, edge deployment, and real-world application value. Overall, this work argues that embodied operators should be optimized and evaluated as holistic deployable components rather than isolated neural networks, providing a foundation for reusable, scalable, and verifiable embodied intelligence systems.  \nDate: July 7, 2026  \nCorrespondence: Junwu Xiong [xiongjunwu.1@jd.com](xiongjunwu.1@jd.com)  \n1 Introduction  \nEmbodied intelligence aims to enable physical or simulated robotic agents to perceive their environments, understand task contexts, and perform interactions through embodied actuation. Recent advances in visionlanguage-action models, robot learning datasets, simulation platforms, and open-source robot learning systems have substantially enhanced the capabilities of embodied agents [1–8] . However, high-quality embodied intelligence systems cannot rely solely on end-to-end policy models. In practical pipelines for data collection, demonstration understanding, scene reconstruction, robot learning, task decision-making, and robot execution, numerous reusable functional modules are required to transform raw multimodal signals and intermediate system states into structured representations, decisions, and executable outputs. In this work, we refer to such reusable functional modules as embodied operators.  \nDifferent from low-level tensor computation primitives, an embodied operator is a deployable functional unit [9] with explicit task semantics, standardized input-output interfaces, and measurable performance characteristics [10] . In embodied data collection and robot execution, these operators play a central role in converting RGB images, depth observations, videos, point clouds, robot states, language instructions, and human demonstrations into usable annotations, representations, plans, and control signals. Typical outputs include hand bounding boxes, hand masks, object poses, camera trajectories, depth maps, reconstructed scenes, hand keypoints, hand motion trajectories, action proposals, grasp poses, collision-free trajectories, contr","cbCaioCx0V3znnoy","https://ap.wps.com/l/cbCaioCx0V3znnoy","pdf",610693,"English","# Introduction\n## Definition and scope of embodied operators\n## Operator taxonomy (five categories)\n## Multi-dimensional benchmarking framework\n## Operator acceleration and open challenges","[{\"question\":\"What problem do embodied operators address in embodied intelligence pipelines?\",\"answer\":\"They provide reusable functional modules with explicit task semantics and standardized input-output interfaces that transform raw multimodal signals and intermediate system states into structured representations, decisions, and executable outputs.\"},{\"question\":\"How are embodied operators categorized in this work?\",\"answer\":\"They are organized into five major categories: detection and segmentation, spatial localization and 3D understanding, hand motion recovery, embodied foundation models and task-decision, and planning/control/system support operators.\"},{\"question\":\"How does the proposed benchmark evaluate embodied operators compared with conventional evaluation?\",\"answer\":\"It extends beyond single-model correctness to consider end-to-end efficiency, resource usage, temporal stability, portability, interface compatibility, deployment reliability, and downstream task utility.\"}]","Embodied Operators and Benchmarking - Toward Reusable and Deployable Embodied Intelligence Systems | PDF",1789977855,76]