Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Jiahan, Jiang, Muqing, Dai, Nanru, Lu, Taiming, Uzunoglu, Arda, Zhang, Shunchi, Wei, Yana, Wang, Jiahao, Patel, Vishal M., Liang, Paul Pu, Khashabi, Daniel, Peng, Cheng, Chellappa, Rama, Shu, Tianmin, Yuille, Alan, Du, Yilun, Chen, Jieneng
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2510.18135
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909860735483904
author Zhang, Jiahan
Jiang, Muqing
Dai, Nanru
Lu, Taiming
Uzunoglu, Arda
Zhang, Shunchi
Wei, Yana
Wang, Jiahao
Patel, Vishal M.
Liang, Paul Pu
Khashabi, Daniel
Peng, Cheng
Chellappa, Rama
Shu, Tianmin
Yuille, Alan
Du, Yilun
Chen, Jieneng
author_facet Zhang, Jiahan
Jiang, Muqing
Dai, Nanru
Lu, Taiming
Uzunoglu, Arda
Zhang, Shunchi
Wei, Yana
Wang, Jiahao
Patel, Vishal M.
Liang, Paul Pu
Khashabi, Daniel
Peng, Cheng
Chellappa, Rama
Shu, Tianmin
Yuille, Alan
Du, Yilun
Chen, Jieneng
contents Generative world models (WMs) can now simulate worlds with striking visual realism, which naturally raises the question of whether they can endow embodied agents with predictive perception for decision making. Progress on this question has been limited by fragmented evaluation: most existing benchmarks adopt open-loop protocols that emphasize visual quality in isolation, leaving the core issue of embodied utility unresolved, i.e., do WMs actually help agents succeed at embodied tasks? To address this gap, we introduce World-in-World, the first open platform that benchmarks WMs in a closed-loop world that mirrors real agent-environment interactions. World-in-World provides a unified online planning strategy and a standardized action API, enabling heterogeneous WMs for decision making. We curate four closed-loop environments that rigorously evaluate diverse WMs, prioritize task success as the primary metric, and move beyond the common focus on visual quality; we also present the first data scaling law for world models in embodied settings. Our study uncovers three surprises: (1) visual quality alone does not guarantee task success, controllability matters more; (2) scaling post-training with action-observation data is more effective than upgrading the pretrained video generators; and (3) allocating more inference-time compute allows WMs to substantially improve closed-loop performance.
format Preprint
id arxiv_https___arxiv_org_abs_2510_18135
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle World-in-World: World Models in a Closed-Loop World
Zhang, Jiahan
Jiang, Muqing
Dai, Nanru
Lu, Taiming
Uzunoglu, Arda
Zhang, Shunchi
Wei, Yana
Wang, Jiahao
Patel, Vishal M.
Liang, Paul Pu
Khashabi, Daniel
Peng, Cheng
Chellappa, Rama
Shu, Tianmin
Yuille, Alan
Du, Yilun
Chen, Jieneng
Computer Vision and Pattern Recognition
Generative world models (WMs) can now simulate worlds with striking visual realism, which naturally raises the question of whether they can endow embodied agents with predictive perception for decision making. Progress on this question has been limited by fragmented evaluation: most existing benchmarks adopt open-loop protocols that emphasize visual quality in isolation, leaving the core issue of embodied utility unresolved, i.e., do WMs actually help agents succeed at embodied tasks? To address this gap, we introduce World-in-World, the first open platform that benchmarks WMs in a closed-loop world that mirrors real agent-environment interactions. World-in-World provides a unified online planning strategy and a standardized action API, enabling heterogeneous WMs for decision making. We curate four closed-loop environments that rigorously evaluate diverse WMs, prioritize task success as the primary metric, and move beyond the common focus on visual quality; we also present the first data scaling law for world models in embodied settings. Our study uncovers three surprises: (1) visual quality alone does not guarantee task success, controllability matters more; (2) scaling post-training with action-observation data is more effective than upgrading the pretrained video generators; and (3) allocating more inference-time compute allows WMs to substantially improve closed-loop performance.
title World-in-World: World Models in a Closed-Loop World
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.18135