[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84182-en":3,"doc-seo-84182-105":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},84182,1374391974468,"Eden","https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0",8,"Research & Report","GeoProp Grounding Robot State in Vision for Generalist Manipulation","Proprioception is central to robotic manipulation, yet common sensor-fusion approaches encode it as an unaligned global vector, leaving the robot’s 3D kinematics disconnected from 2D visual feature maps. This decoupling forces policies to learn state-to-vision alignment implicitly and can underperform vision-only baselines. GeoProp is introduced as a lightweight plug-and-play adapter: it projects robot state onto the image plane, samples localized visual features to form a grounded state token, and modulates visual tokens via FiLM, with short-horizon predictive sampling for motion intent. Across 67 tasks, GeoProp improves diffusion-based and π0 policies with minimal parameter overhead.","arXiv :2607 .07 10 1v 1 [ cs .RO] 8 Jul 2026  \nGeoProp: Grounding Robot State in Vision for Generalist Manipulation  \nGuoyang Zhao 1 ,∗ , Quanhao Qian2 ,3 ,∗ , Gongjie Zhang4 , Wenhao Li5 , Jiuniu Wang2 ,3 , Xiaowei Lu2 ,3 , Deli Zhao2 ,3 , Ran Xu2 ,3 , B  \n1Tongji University 2DAMO Academy, Alibaba Group 3HuPan Lab  \n4Alibaba Group 5Nanyang Technological University  \n∗Equal contribution. BCorresponding author.  \nAbstract: Proprioception is fundamental to robotic manipulation, yet standard fusion methods often treat it as an isolated vector lacking explicit alignment with visual tokens. Without a direct correspondence between 3D kinematics and 2D feature maps, manipulation policies struggle to ground the robot’s state within the scene, frequently underperforming even vision-only baselines. To address this, we introduce GeoProp, a lightweight, plug-and-play adapter that aligns proprioception with vision through explicit geometric grounding and spatial feature sampling. GeoProp projects the robot state onto the image plane to sample localized visual features, constructing a grounded state token. It then injects state-derived spatial priors into the corresponding visual features via FiLM modulation. To capture motion intent, GeoProp further samples features at a short-horizon predicted coordinate derived from recent kinematics, providing look-ahead visual context. Across 67 tasks, GeoProp improves Diffusion Policy by 8.7% on 63 simulation tasks and π0 by 4.0% on the RoboTwin subset, and yields a 10.6% average gain across both policy families in the real world, while adding only 2– 3% to the parameter count. These results demonstrate that GeoProp is a simple yet high-impact inductive bias for generalist embodied policies. Project page: [https://alibaba-damo-academy.github.io/GeoProp/](https://alibaba-damo-academy.github.io/GeoProp/) .  \nKeywords: Robot Perception, Proprioception, Visual Grounding  \n1 Introduction  \nDespite the diversification of robot learning architectures—spanning diffusion-based controllers [1, 2, 3], transformer-based action predictors [4, 5], and large-scale vision–language–action (VLA) systems [6, 7, 8, 9]—a persistent representational limitation remains: the structural decoupling between high-dimensional visual observations and low-dimensional proprioceptive feedback. Current mainstream frameworks encode proprioception asa global, ungrounded state embedding and fuse it with vision either through simple concatenation [10, 4, 11, 12] or through more expressive cross-attention [9, 13] . Both designs typically lack an explicit state-vision correspondence and require the model to learn 3D kinematics-to-2D feature alignment implicitly, overlooking the robot’s intrinsic geometric relationship with the visual scene. Our empirical analysis shows that ungrounded  \nFigure 1: Proprioceptive-to-image attention in π0 : GeoProp concentrates attention on the gripper and manipulated objects, while vanilla attention is diffuse.  \nproprioception can be counterproductive: without explicit alignment, the state vector may introduce spurious correlations that cause the model to underperform vision-only alternatives.  \nWe propose GeoProp, a lightweight adapter that bridges the gap between 3D kinematics and 2D vision by transforming the robotic state into a localized geometric modulation within the 2D visual feature map. Instead of treating proprioception as an ungrounded auxiliary input, GeoProp projects the end-effector pose onto the image plane and uses the corresponding localized visual features asthe state-aligned visual feature of the robot. This enables the robot’s configuration to inherit scene semantics within the same visual latent space, promoting modality-consistent fusion, as visualized by the attention heatmaps in Fig. 1 and Appendix B.3 . In the vanilla π0 [9] model, state embeddingsoften fail to produce spatially coherent attention and yield diffuse, background-driven patterns. In contrast, GeoProp tightly couples pr","cbCaintbsnvNdPj3","https://ap.wps.com/l/cbCaintbsnvNdPj3","pdf",11786566,1,21,"English","en",105,"# Introduction\n## Problem: Decoupling of Visual Observations and Proprioceptive Feedback\n## Proposed Method: GeoProp\n## Evaluation Results and Contributions","[{\"question\":\"What problem does GeoProp address in robotic manipulation learning?\",\"answer\":\"GeoProp targets the representational decoupling between high-dimensional vision and low-dimensional proprioceptive feedback. Standard fusion treats proprioception as an ungrounded vector, lacking explicit state-to-vision correspondence.\"},{\"question\":\"How does GeoProp align robot state with visual features?\",\"answer\":\"GeoProp projects the robot end-effector pose onto the image plane, samples localized visual features near the projected location to build a grounded state token, and uses FiLM modulation to inject state-derived spatial priors into visual tokens.\"},{\"question\":\"How effective is GeoProp, and what is its computational cost?\",\"answer\":\"On 67 tasks, GeoProp improves diffusion-policy performance by 8.7% on 63 simulation tasks and π0 by 4.0% on the RoboTwin subset, with a 10.6% average gain in real-world tests. It adds only about 2–3% to parameter count.\"}]",1784193722,53,{"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},"geoprop-grounding-robot-state-in-vision-for-generalist-manipulation","",{"@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/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/geoprop-grounding-robot-state-in-vision-for-generalist-manipulation/84182/",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 GeoProp address in robotic manipulation learning?","Question",{"text":75,"@type":76},"GeoProp targets the representational decoupling between high-dimensional vision and low-dimensional proprioceptive feedback. Standard fusion treats proprioception as an ungrounded vector, lacking explicit state-to-vision correspondence.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does GeoProp align robot state with visual features?",{"text":80,"@type":76},"GeoProp projects the robot end-effector pose onto the image plane, samples localized visual features near the projected location to build a grounded state token, and uses FiLM modulation to inject state-derived spatial priors into visual tokens.",{"name":82,"@type":73,"acceptedAnswer":83},"How effective is GeoProp, and what is its computational cost?",{"text":84,"@type":76},"On 67 tasks, GeoProp improves diffusion-policy performance by 8.7% on 63 simulation tasks and π0 by 4.0% on the RoboTwin subset, with a 10.6% average gain in real-world tests. It adds only about 2–3% to parameter count.","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},[93,97,101,105,110,115,120,123,128,131,135],{"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":111,"doc_module":4,"doc_module_name":45,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":45,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":45,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]