World models infer and predict real-world dynamics by modeling the external environment, and have become a cornerstone of embodied artificial intelligence. They have powered recent progress in decision-making and planning for interacting agents. This workshop aims to bring together researchers working at the intersection of generative modeling, reinforcement learning, computer vision, and robotics to explore the next generation of embodied world models—models that enable agents to understand, predict, and interact with the world through learned models. By focusing on embodiment and decision-making, this workshop seeks to advance world models beyond passive prediction, toward active, goal-driven interaction with the physical and virtual world. By emphasizing embodiment and decision-making, we aim to move beyond passive sequence prediction toward goal-directed interaction with both physical and simulated worlds.
Topics of Interest
We welcome contributions that advance theoretical foundations, algorithmic innovations, or real-world applications of world models. Topics of interest include (but are not limited to):
Model-based reinforcement learning and long-horizon planning.
Investigating how world models can benefit model-based reinforcement learning with a focus on sample efficiency, performance, and scalability. Particular attention is given to long-horizon planning, which requires the agent to reason over extended sequences of actions, anticipate delayed outcomes, and maintain coherent strategies across temporally distant states and goals, often under uncertainty and limited feedback.
Aligning simulation and real-world physics for robot learning.
Investigating how to bridge the gap between simulated and real-world physics to enhance robot learning. This includes using generative models to improve perception, planning, and control by capturing physical dynamics more accurately, modeling uncertainty and feedback effects, and learning diffusion-based policies that transfer robustly from simulation to the real world.
Interactive scene generation and downstream tasks.
Building models that generate physically plausible and semantically coherent interactive video simulations. Focus areas include action-conditioned scene synthesis, controllable simulation of agent-environment dynamics, and the development of evaluation techniques and benchmarks that assess video fidelity, temporal consistency, and task-relevant controllability for downstream applications such as planning and policy learning.
Video-language-action (VLA) models and leveraging the world knowledge encoded in large language models (LLMs).
Studying large-scale pretrained models that unify video, language, and action representations to support robust and generalizable policy learning. Core areas include curating diverse multi-modal datasets, improving cross-modal alignment, developing parameter-efficient fine-tuning methods, and enabling agents to follow complex, language-guided instructions in both simulated and real-world settings. We also explore how the structured and unstructured world knowledge embedded in large language models can be exploited to guide agents’ decision-making.
Applications in broader domains, such as open-world video games and autonomous driving.
Extending world models to embodied agents in both real-world environments and high-fidelity simulators. Key topics include integrating perception with control, sim-to-real transfer, continual learning and adaptation, and deploying agents in open-ended tasks such as Minecraft, autonomous driving, and interactive real-world scenarios.
Call for Papers
We invite submissions of original research papers related to
building physically plausible world models.
Submission Types:
Opinion Papers (max 4 pages with unlimited references, NeurIPS format)
- Opinion papers that propose new visions, future directions, or highlight challenges and opportunities in embodied world models for decision-making, without requiring extensive experimental results.
Research Papers (4 to 9 pages with unlimited references and appendices, NeurIPS format) - For original
research contributions.
Papers are non-archival - we welcome submissions that have been submitted to or accepted by other venues.
Papers submitted to the workshop will be reviewed in a double-blind process.
For opinion papers, the title should state the opinion and start with "Opinion:", such as "Opinion: Large Language Models Should Not Replace Peer Review in Scientific Publishing".
Papers already accepted to NeurIPS 2025 will undergo an expedited
review process primarily evaluating their relevance to the
workshop themes.
All accepted papers will be presented in a poster session
Important Dates:
Submission Deadline: September 1, 2025 11:59PM UTC-0 New Submission Deadline: September 3, 2025 11:59PM UTC-0
Notification of Acceptance: September 17, 2025 UTC-0
New Notification of Acceptance: September 20, 2025 UTC-0
Camera Ready Deadline: October 26, 2025 11:59PM UTC-0
09:30 - 10:00Keynote #2Sanja Fidler (University of Toronto & NVIDIA) - Towards World Models for Autonomous Driving
10:00 - 10:30Keynote #3Nicklas Hansen (UC San Diego) - Massively Multitask World Models for Continuous Control
10:30 - 11:00Keynote #4Chelsea Finn (Stanford University & Physical Intelligence) - Developing Long-Term Autonomy
11:00 - 11:30Keynote #5Peter Stone (The University of Texas at Austin & Sony AI) - Beyond Sim2Real: Leveraging Simulation in Support of Embodied World Models for Real-World Robot Learning
11:30 - 12:00Keynote #6Glen Berseth (Mila) - Using Foundational Models for Embodied Control
12:00 - 13:00 Lunch Break + Poster Session II
13:00 - 13:15Industry DemoGianluca Corrado (Wayve) &
Lorenzo Bertoni (Wayve) - Scaling World Models to Power Evaluation and Validation
13:15 - 13:45Keynote #7John Langford (Microsoft Research New York) - Next-Latent Prediction Transformers Learn Compact World Models
13:45 - 14:15Keynote #8Yilun Du (Harvard) - Building Intelligent Robots with World Models
14:15 - 14:45Keynote #9Philip J. Ball (DeepMind) - Genie 3: A new frontier for world models
15:35 - 16:05Keynote #10Jianlan Luo (AgiBot) - World Model Powered Robotic Manipulation
16:05 - 16:35Keynote #11Pablo Samuel Castro (DeepMind & Mila) - Automated Reward Machines via Foundation Models for Compositional Reinforcement Learning
16:35 - 16:41Oral:Sandeep Routray (CMU) - ViPRA: Video Prediction for Robot Actions
(Oral) Robotic World Model: A Neural Network Simulator for Robust Policy Optimization in Robotics[Video] Chenhao Li · Andreas Krause · Marco Hutter
Divide and Merge: Motion and Semantic Learning in End-to-End Autonomous Driving Yinzhe Shen · Omer Sahin Tas · Kaiwen Wang · Royden Wagner · Christoph Stiller
NinA: Normalizing Flows in Action. Training VLA Models with Normalizing Flows Denis Tarasov · Alexander Nikulin · Ilya Zisman · Albina Klepach · Nikita Lyubaykin · Andrei Polubarov · Alexander Derevyagin · Vladislav Kurenkov
Communicating Plans, Not Percepts: Scalable Multi-Agent Coordination with Embodied World Models Brennen Hill · Mant Wei · Jishnu Anandh Thangavel
Coupled Distributional Random Expert Distillation for World Model Online Imitation Learning Shangzhe Li · Zhiao Huang · Hao Su
LLM-Guided Probabilistic Program Induction for POMDP Model Estimation Aidan Curtis · Hao Tang · Thiago Veloso · Kevin Ellis · Josh Tenenbaum · Tomás Lozano-Pérez · Leslie Kaelbling
Exploring exploration with foundation agents in interactive environments Daniel Sawyer · Nan Rosemary Ke · Hubert Soyer · Martin Engelcke · John Reid · David Reichert · Drew Hudson · Alexander Lerchner · Danilo Jimenez Rezende · Timothy Lillicrap · Michael Mozer · Jane Wang
Bridging the Sim-to-Real Gap in Humanoid Dynamics via Learned Nonlinear Operators Jieming Cui · Zhenghao Qi · Yutang Lin · Yifei Zhao · Yuntian Hu · Lei Huang · Shuang Qiu · Rita Zhang · Bin He · Yixin Zhu
FalconWing: An Ultra-Light Fixed-Wing Platform for Indoor Aerial Applications Yan Miao · Will Shen · Hang Cui · Sayan Mitra
Decoupled Planning and Execution with LLM-Driven World Models for Efficient Task Planning Guoqing Ma
RDAR: Reward-Driven Agent Relevance Estimation for Autonomous Driving Carlo Bosio · Greg Woelki · Noureldin Hendy · Nick Roy · Byungsoo Kim
Ada-Diffuser: Latent-Aware Adaptive Diffusion for Decision-Making Fan Feng · Selena Ge · Minghao Fu · Zijian Li · Yujia Zheng · Zeyu Tang · Yingyao Hu · Biwei Huang · Kun Zhang
The Physical Basis of Prediction: World Model Formation in Neural Organoids via an LLM-Generated Curriculum Brennen Hill
ScenePhys — Controllable Physics Videos for World-Model Evaluation Arshia Hemmat · Ernad Aghahosseini · Alireza Nasri · Mohammad Hossein Shaker Ardakani · Amirmasoud Rismanchian · Ali Mamanpoosh · Afsaneh Fatemi
A Smooth Sea Never Made a Skilled $\texttt{SAILOR}$: Robust Imitation via Learning to Search Arnav Kumar Jain · Vibhakar Mohta · Subin Kim · Atiksh Bhardwaj · Juntao Ren · Yunhai Feng · Sanjiban Choudhury · Gokul Swamy
In-Context Policy Iteration for Dynamic Manipulation Mark Van der Merwe · Devesh Jha
Learning to Focus: Prioritizing Informative Histories with Structured Attention Mechanisms in Partially Observable Reinforcement Learning Daniel De Dios Allegue · Jinke He · Frans Oliehoek
PolicyGRID: Acting to Understand, Understanding to Act Taqiya Ehsan · Shuren Xia · Jorge Ortiz
Sim-to-Real Contact-Rich Pivoting via Optimization-Guided RL with Vision and Touch Yuki Shirai · Kei Ota · Devesh Jha · Diego Romeres
Opinion: Towards Unified Expressive Policy Optimization for Robust Robot Learning Haidong Huang · Haiyue Zhu · Jiayi Song · Xixin Zhao · Yaohua Zhou · Jiayi Zhang · Yuze Zhai · Xiaocong Li
Opinion: Learning Intuitive Physics Requires More Than Visual Data Ellen Su · Solim LeGris · Todd Gureckis · Mengye Ren
Opinion: A Unified World Model is the cornerstone for integrating perception, reasoning, and decision-making in embodied AI Yipeng Xu
Geosteering Through the Lens of Decision Transformers: Toward Embodied Sequence Decision-Making Hibat Errahmen DJECTA
Avi: A 3D Vision-Language Action Model Architecture generating Action from Volumetric Inference Harris Song · Long Le
Hierarchical Task Environments as the Next Frontier for Embodied World Models in Robot Soccer Brennen Hill
Plan Verification for LLM-Based Embodied Task Completion Agents Ananth Harharan · Vardhan Dongre · Dilek Tur · Gokhan Tur
WHALE: Towards Generalizable and Scalable World Models for Embodied Decision-making Zhilong Zhang · Ruifeng Chen · Junyin Ye · Yihao Sun · Haoxiang Ren · Xinghua Du · Pengyuan Wang · Jing-Cheng Pang · Kaiyuan Li · Tian-Shuo Liu · Haoxin Lin · Yang Yu · Zhi-Hua Zhou
OpenGVL - Benchmarking Visual Temporal Progress for Data Curation Pawel Budzianowski · Emilia Wiśnios · Gracjan Góral · Igor Kulakov · Viktor Petrenko · Krzysztof Walas
Foundation Models as World Models: A Foundational Study in Text-Based GridWorlds Remo Sasso · Michelangelo Conserva · Dominik Jeurissen · Paulo Rauber
Steering Diffusion Policies with Value-Guided Denoising Hanming Ye
SPUR: Scaling Reward Learning from Human Demonstrations Anthony Liang · Yigit Korkmaz · Jiahui Zhang · Jesse Zhang · Abrar Anwar · Sid Kaushik · Yufei Wang · Yu Xiang · David Held · Dieter Fox · Abhishek Gupta · Stephen Tu · Erdem Bıyık
Stable Planning through Aligned Representations in Model-Based Reinforcement Learning Misagh Soltani · Forest Agostinelli
SpatialThinker: Reinforcing 3D Reasoning in Multimodal LLMs via Spatial Rewards Hunar Batra · Haoqin Tu · Hardy Chen · Yuanze Lin · Chang Xie · Ronald Clark
How Foundational Skills Influence VLM-based Embodied Agents: A Native Perspective Bo Peng · Pi Bu · Keyu Pan · Xinrun Xu · Miao Chen · Yang Du · Lin Li · Jun Song · Tong Xu · Bo Zheng
Beyond Experience: Fictive Learning as an Inherent Advantage of World Models Jianning Chen · Masakazu Taira · Kenji Doya
Opinion: Small VLAs Self-Learn Consistency Francesco Capuano · Adil Zouitine · Michel Aractingi
HDFlow: Hierarchical Diffusion-Flow Planning for Long-horizon Robotic Assembly Gireesh Nandiraju · Yuanliang Ju · Chaoyi Xu · He Wang
Mobile Manipulation with Active Inference for Long-Horizon Rearrangement Tasks Corrado Pezzato · Ozan Catal · Toon Van de Maele · Riddhi Pitliya · Tim Verbelen
Towards Fine-tuning a Small Vision-Language Model for Aerial Navigation Hakob Tamazyan · Narek Nurijanyan · Boris Martirosyan · Hrant Khachatrian
CRISP: Contact-guided Real2Sim from Monocular Video with Planar Scene Primitives Zihan Wang · Jiashun Wang · Jeff Tan · Yiwen Zhao · Jessica Hodgins · Shubham Tulsiani · Deva Ramanan
Abstract Sim2Real through Approximate Information States Yunfu Deng · Josiah Hanna
Opinion: How Can Causal AI Benefit World Models? Qiuling Pan · Hong Zhou · Zhouchen Lin
ROPES: Robotic Pose Estimation via Score-based Causal Representation Learning Pranamya Kulkarni · Puranjay Datta · Emre Acartürk · Burak Varıcı· Karthikeyan Shanmugam · Ali Ahmed
FLAM: Scaling Latent Action Models with Factorization Chang Shi · Zizhao Wang · Jiaheng Hu · Roberto Martín-Martín · Peter Stone
Improvisational Reasoning with Vision-Language Models for Grounded Procedural Planning Masudur Rahman · Yupeng Zhuo · Juan Wachs
Vision-Language Reasoning for Burn Depth Assessment with Structured Diagnostic Hypotheses Masudur Rahman · Mohamed Masry · Kristo Nuutila · Gayle Gordillo · Juan Wachs
