Hugging Face Daily Papers · June 26, 2026 · 3 min read

In-Context World Modeling for Robotic Control

Mirrored from Hugging Face Daily Papers for archival readability. Support the source by reading on the original site.

Like Read original ↗

In-Context World Modeling for Robotic Control\n","updatedAt":"2026-06-26T02:48:36.684Z","author":{"_id":"64c3c631e77ea9f28111172a","avatarUrl":"/avatars/495dbb73b69c399bae780da3118e332f.svg","fullname":"Siyin Wang (SII)","name":"sinwang","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":17,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.5499311685562134},"editors":["sinwang"],"editorAvatarUrls":["/avatars/495dbb73b69c399bae780da3118e332f.svg"],"reactions":[],"isReport":false}},{"id":"6a40b5838fbfc742ff9e3590","author":{"_id":"679644efaae193052def45c0","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/UZZzkjp4IFcz2eeTSa41e.png","fullname":"Wen JunTing","name":"Wenwen555","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"createdAt":"2026-06-28T05:47:47.000Z","type":"comment","data":{"edited":false,"hidden":false,"latest":{"raw":"@librarian-bot","html":"<a href=\"/librarian-bot\">@librarian-bot</a> \n","updatedAt":"2026-06-28T05:47:47.215Z","author":{"_id":"679644efaae193052def45c0","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/UZZzkjp4IFcz2eeTSa41e.png","fullname":"Wen JunTing","name":"Wenwen555","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7558995485305786},"editors":["Wenwen555"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/UZZzkjp4IFcz2eeTSa41e.png"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.26025","authors":[{"_id":"6a3de83a3b43e283349ec1a1","user":{"_id":"64c3c631e77ea9f28111172a","avatarUrl":"/avatars/495dbb73b69c399bae780da3118e332f.svg","isPro":false,"fullname":"Siyin Wang (SII)","user":"sinwang","type":"user","name":"sinwang"},"name":"Siyin Wang","status":"claimed_verified","statusLastChangedAt":"2026-06-27T15:23:31.616Z","hidden":false},{"_id":"6a3de83a3b43e283349ec1a2","name":"Junhao Shi","hidden":false},{"_id":"6a3de83a3b43e283349ec1a3","name":"Senyu Fei","hidden":false},{"_id":"6a3de83a3b43e283349ec1a4","name":"Zhaoyang Fu","hidden":false},{"_id":"6a3de83a3b43e283349ec1a5","name":"Li Ji","hidden":false},{"_id":"6a3de83a3b43e283349ec1a6","name":"Jingjing Gong","hidden":false},{"_id":"6a3de83a3b43e283349ec1a7","name":"Xipeng Qiu","hidden":false}],"publishedAt":"2026-06-25T00:00:00.000Z","submittedOnDailyAt":"2026-06-26T00:00:00.000Z","title":"In-Context World Modeling for Robotic Control","submittedOnDailyBy":{"_id":"64c3c631e77ea9f28111172a","avatarUrl":"/avatars/495dbb73b69c399bae780da3118e332f.svg","isPro":false,"fullname":"Siyin Wang (SII)","user":"sinwang","type":"user","name":"sinwang"},"summary":"Modern Vision-Language-Action (VLA) models often fail to generalize to novel setups, such as altered camera viewpoints or robot morphologies, because they are typically conditioned only on current observations and language instructions. By ignoring the underlying system configuration as a variable, these models implicitly assume a fixed execution context encountered during training, necessitating data-intensive fine-tuning for any new environment. In this work, we introduce In-Context World Modeling (ICWM), a framework that treats system identification as an in-context adaptation problem. ICWM enables robot policies to autonomously infer essential system variables from a short history of self-generated, task-agnostic interactions. Unlike traditional In-Context Learning that uses demonstrations to specify what task to perform, ICWM leverages the context window to understand how the system operates. By processing these interactions before task execution, the model implicitly captures the world dynamics of the current system, enabling adaptation to novel configurations without parameter updates. Extensive experiments in simulation and on real-world robot platforms demonstrate that ICWM significantly outperforms standard VLA baselines on novel camera viewpoints.","upvotes":45,"discussionId":"6a3de83b3b43e283349ec1a8","ai_summary":"ICWM enables robot policies to infer system variables from self-generated interactions, allowing adaptation to novel configurations without parameter updates by treating system identification as an in-context adaptation problem.","ai_keywords":["Vision-Language-Action models","in-context adaptation","system identification","robot policies","task-agnostic interactions","world modeling","novel configurations","parameter updates","simulation","real-world robot platforms"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","organization":{"_id":"613b0dee83ec35d460684607","name":"OpenMOSS-Team","fullname":"OpenMOSS","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/61457b8deff2c9fdb4de4988/N5b9663zQ4uq5_OTNlnmw.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"64c3c631e77ea9f28111172a","avatarUrl":"/avatars/495dbb73b69c399bae780da3118e332f.svg","isPro":false,"fullname":"Siyin Wang (SII)","user":"sinwang","type":"user"},{"_id":"67a8b856aaa85bdabef9f2c3","avatarUrl":"/avatars/fb01d274593cbf320d083e9bcdd62617.svg","isPro":false,"fullname":"yu","user":"shengjiexy","type":"user"},{"_id":"6270324ebecab9e2dcf245de","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6270324ebecab9e2dcf245de/cMbtWSasyNlYc9hvsEEzt.jpeg","isPro":false,"fullname":"Kye Gomez","user":"kye","type":"user"},{"_id":"6401a0c14a5c92eccfe333c2","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6401a0c14a5c92eccfe333c2/-fv3jekUxMNrE058phGpA.png","isPro":false,"fullname":"Ha-Yeong Choi","user":"Ha0","type":"user"},{"_id":"677272184d148b904333e874","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/5dUau7gxLk4Wm1TiiJJri.jpeg","isPro":false,"fullname":"Efstathios Karypidis","user":"Sta8is","type":"user"},{"_id":"63c1699e40a26dd2db32400d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63c1699e40a26dd2db32400d/3N0-Zp8igv8-52mXAdiiq.jpeg","isPro":false,"fullname":"Chroma","user":"Chroma111","type":"user"},{"_id":"6a2da6c8ca070ee12c6e396c","avatarUrl":"/avatars/0355287dcabaa67dbc7f0b10b87451f9.svg","isPro":false,"fullname":"Joe Mama","user":"JoeMama123123123","type":"user"},{"_id":"69f0bb9a53592156859aab90","avatarUrl":"/avatars/122aeb140c584b7842c50ae693c2a27e.svg","isPro":false,"fullname":"mini09999","user":"mini09999","type":"user"},{"_id":"64f033ef82c6eea604c4da8b","avatarUrl":"/avatars/51b93fea7fd68b4274ee03701245dcca.svg","isPro":false,"fullname":"Xiaoran Liu (SII)","user":"SII-xrliu","type":"user"},{"_id":"67f5e63688b2c5303ab5be7a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/67f5e63688b2c5303ab5be7a/QSH8-QZH6l3KradXqNxJT.png","isPro":false,"fullname":"Chengxuan Qian","user":"Raymond-Qiancx","type":"user"},{"_id":"696da0962b3e2d9587d0b35d","avatarUrl":"/avatars/4f6c177ad51fb687ca1be75d18f6f5d6.svg","isPro":false,"fullname":"mini","user":"mini0999","type":"user"},{"_id":"687f904b5dd92a729b0c6b65","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/687f904b5dd92a729b0c6b65/EUU_RhuKjct46kJkDbNnP.jpeg","isPro":false,"fullname":"Yikang Zhou","user":"Kevin517","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":2,"organization":{"_id":"613b0dee83ec35d460684607","name":"OpenMOSS-Team","fullname":"OpenMOSS","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/61457b8deff2c9fdb4de4988/N5b9663zQ4uq5_OTNlnmw.png"},"query":{}}">

Papers

arxiv:2606.26025

In-Context World Modeling for Robotic Control

Published on Jun 25

· Submitted by

Siyin Wang (SII) on Jun 26

#2 Paper of the day

OpenMOSS

Upvote

Authors:

Siyin Wang ,

Abstract

ICWM enables robot policies to infer system variables from self-generated interactions, allowing adaptation to novel configurations without parameter updates by treating system identification as an in-context adaptation problem.

Generated by Qwen/Qwen2.5-Coder-32B-Instruct

Modern Vision-Language-Action (VLA) models often fail to generalize to novel setups, such as altered camera viewpoints or robot morphologies, because they are typically conditioned only on current observations and language instructions. By ignoring the underlying system configuration as a variable, these models implicitly assume a fixed execution context encountered during training, necessitating data-intensive fine-tuning for any new environment. In this work, we introduce In-Context World Modeling (ICWM), a framework that treats system identification as an in-context adaptation problem. ICWM enables robot policies to autonomously infer essential system variables from a short history of self-generated, task-agnostic interactions. Unlike traditional In-Context Learning that uses demonstrations to specify what task to perform, ICWM leverages the context window to understand how the system operates. By processing these interactions before task execution, the model implicitly captures the world dynamics of the current system, enabling adaptation to novel configurations without parameter updates. Extensive experiments in simulation and on real-world robot platforms demonstrate that ICWM significantly outperforms standard VLA baselines on novel camera viewpoints.