Papers

A curated canon of the literature this course teaches from. Filter by concept (the part of the world-model stack a paper belongs to) and by level:

• Foundation — the classics everything else assumes (Kalman 1960, VAE, PointNet, ORB-SLAM…)
• Must Read — the papers you should actually read in full to complete a track
• Advanced — deep dives and recent frontier work for after the core modules

Concept

Level

85 papers

• A Tutorial on World Models and Physical AII.-S. Oh · 2026 · arXiv
  FoundationSurvey
• Nano World Models: A Minimalist Implementation of Future Video PredictionS. Huang, P. Kaushik, M. Chen, et al. · 2026 · arXiv
  FoundationVideo
• From Masks to Worlds: A Hitchhiker's Guide to World ModelsJ. Bai, Y. Lei, H. Wu, et al. · 2025 · arXiv
  FoundationSurvey
• Understanding World or Predicting Future? A Comprehensive Survey of World ModelsJ. Ding, Y. Zhang, Y. Shang, et al. · 2024 · arXiv / ACM Computing Surveys
  FoundationSurvey
• Masked Autoencoders Are Scalable Vision LearnersK. He, X. Chen, S. Xie, Y. Li, P. Dollár, R. Girshick · 2022 · CVPR
  FoundationRepresentation
• Volume Rendering Digest (for NeRF)A. Tagliasacchi, B. Mildenhall · 2022 · arXiv
  Foundation3D
• Towards Causal Representation LearningB. Schölkopf, F. Locatello, S. Bauer, et al. · 2021 · Proceedings of the IEEE
  FoundationTheory
• Occupancy Networks: Learning 3D Reconstruction in Function SpaceL. Mescheder, M. Oechsle, M. Niemeyer, S. Nowozin, A. Geiger · 2019 · CVPR
  Foundation3D
• World ModelsD. Ha, J. Schmidhuber · 2018 · NeurIPS Workshop / arXiv
  FoundationDynamics
• Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models (PETS)K. Chua, R. Calandra, R. McAllister, S. Levine · 2018 · NeurIPS
  FoundationPlanning
• PointNet: Deep Learning on Point Sets for 3D Classification and SegmentationC. R. Qi, H. Su, K. Mo, L. J. Guibas · 2017 · CVPR
  Foundation3D
• Mastering the Game of Go with Deep Neural Networks and Tree Search (AlphaGo)D. Silver, A. Huang, C. Maddison, et al. · 2016 · Nature
  FoundationPlanning
• The EuRoC Micro Aerial Vehicle DatasetsM. Burri, J. Nikolic, P. Gohl, et al. · 2016 · IJRR
  FoundationEvaluation
• ORB-SLAM: A Versatile and Accurate Monocular SLAM SystemR. Mur-Artal, J. M. M. Montiel, J. D. Tardós · 2015 · IEEE T-RO
  FoundationRobotics
• Auto-Encoding Variational BayesD. P. Kingma, M. Welling · 2014 · ICLR
  FoundationRepresentation
• A New Approach to Linear Filtering and Prediction ProblemsR. E. Kalman · 1960 · Journal of Basic Engineering
  FoundationDynamics
• LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from PixelsL. Maes, Q. Le Lidec, D. Scieur, et al. · 2026 · arXiv
  Must ReadDynamics
• General Agents Contain World ModelsJ. Richens, D. Abel, A. Bellot, et al. · 2025 · ICML
  Must ReadTheory
• What Has a Foundation Model Found? Using Inductive Bias to Probe for World ModelsK. Vafa, P. G. Chang, A. Rambachan, et al. · 2025 · ICML
  Must ReadTheory
• Revisiting Feature Prediction for Learning Visual Representations from Video (V-JEPA)A. Bardes, Q. Garrido, J. Ponce, X. Chen, M. Rabbat, Y. LeCun, M. Assran, N. Ballas · 2024 · ICML
  Must ReadRepresentation
• Genie: Generative Interactive EnvironmentsJ. Bruce, M. Dennis, A. Edwards, et al. · 2024 · ICML
  Must ReadVideo
• Video Generation Models as World Simulators (Sora technical report)OpenAI (T. Brooks, B. Peebles, et al.) · 2024 · OpenAI Technical Report
  Must ReadVideo
• DUSt3R: Geometric 3D Vision Made EasyS. Wang, V. Leroy, Y. Cabon, B. Chidlovskii, J. Revaud · 2024 · CVPR
  Must Read3D
• Octo: An Open-Source Generalist Robot PolicyOcto Model Team (D. Ghosh, H. Walke, et al.) · 2024 · RSS
  Must ReadRobotics
• Evaluating Real-World Robot Manipulation Policies in Simulation (SimplerEnv)X. Li, K. Hsu, J. Gu, et al. · 2024 · CoRL
  Must ReadEvaluation
• Thinking in Space: How Multimodal Large Language Models See, Remember, and Recall Spaces (VSI-Bench)J. Yang, S. Yang, A. W. Gupta, R. Han, L. Fei-Fei, S. Xie · 2024 · arXiv
  Must ReadEvaluation
• Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture (I-JEPA)M. Assran, Q. Duval, I. Misra, P. Bojanowski, P. Vincent, M. Rabbat, Y. LeCun, N. Ballas · 2023 · CVPR
  Must ReadRepresentation
• Mastering Diverse Domains through World Models (DreamerV3)D. Hafner, J. Pasukonis, J. Ba, T. Lillicrap · 2023 · arXiv / Nature (2025)
  Must ReadDynamics
• 3D Gaussian Splatting for Real-Time Radiance Field RenderingB. Kerbl, G. Kopanas, T. Leimkühler, G. Drettakis · 2023 · SIGGRAPH
  Must Read3D
• RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic ControlA. Brohan, N. Brown, J. Carbajal, et al. · 2023 · CoRL
  Must ReadRobotics
• Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Task (Othello-GPT)K. Li, A. K. Hopkins, D. Bau, et al. · 2023 · ICLR
  Must ReadTheory
• Temporal Difference Learning for Model Predictive Control (TD-MPC)N. Hansen, X. Wang, H. Su · 2022 · ICML
  Must ReadPlanning
• Dream to Control: Learning Behaviors by Latent Imagination (DreamerV1)D. Hafner, T. Lillicrap, J. Ba, M. Norouzi · 2020 · ICLR
  Must ReadDynamics
• NeRF: Representing Scenes as Neural Radiance Fields for View SynthesisB. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoorthi, R. Ng · 2020 · ECCV
  Must Read3D
• Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model (MuZero)J. Schrittwieser, I. Antonoglou, T. Hubert, et al. · 2020 · Nature
  Must ReadPlanning
• Learning Latent Dynamics for Planning from Pixels (PlaNet)D. Hafner, T. Lillicrap, I. Fischer, et al. · 2019 · ICML
  Must ReadDynamics
• World Model for Robot Learning: A Comprehensive SurveyB. Hou, G. Li, J. Jia, et al. · 2026 · arXiv
  AdvancedSurvey
• From Observation to Insight: Mechanistic World Models and the Quest for Autonomous DiscoveryI. Posner, A. Lei, B. Schölkopf · 2026 · arXiv
  AdvancedTheory
• World Action Models are Zero-shot Policies (DreamZero)S. Ye, Y. Ge, K. Zheng, et al. · 2026 · arXiv
  AdvancedRobotics
• V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and PlanningM. Assran, A. Bardes, D. Fan, et al. · 2025 · arXiv
  AdvancedRepresentation
• VGGT: Visual Geometry Grounded TransformerJ. Wang, M. Chen, N. Karaev, A. Vedaldi, C. Rupprecht, D. Novotny · 2025 · CVPR
  Advanced3D
• Navigation World ModelsA. Bar, D. Tsur, A. Zentner, et al. · 2025 · CVPR
  AdvancedRobotics
• Cosmos World Foundation Model Platform for Physical AINVIDIA · 2025 · arXiv
  AdvancedVideo
• Do Generative Video Models Understand Physical Principles? (Physics-IQ)S. Motamed, L. Culp, K. Swersky, P. Jaini, R. Geirhos · 2025 · arXiv
  AdvancedEvaluation
• 3D and 4D World Modeling: A SurveyL. Kong, Y. Yang, J. Mei, et al. · 2025 · arXiv
  AdvancedSurvey
• A Comprehensive Survey on World Models for Embodied AIX. Li, X. He, L. Zhang, et al. · 2025 · arXiv
  AdvancedSurvey
• A Survey of World Models for Autonomous DrivingT. Feng, W. Wang, Y. Yang · 2025 · arXiv
  AdvancedSurvey
• When Do Neural Networks Learn World Models?T. Zhang, G. Chen, F. Chen · 2025 · ICML
  AdvancedTheory
• Critique of World ModelE. Xing, M. Deng, J. Hou · 2025 · arXiv
  AdvancedTheory
• LeJEPA: Provable and Scalable Self-Supervised Learning Without the HeuristicsR. Balestriero, Y. LeCun · 2025 · arXiv
  AdvancedRepresentation
• Back to the Features: DINO as a Foundation for Video World ModelsF. Baldassarre, M. Szafraniec, B. Terver, et al. · 2025 · arXiv
  AdvancedVideo
• Intuitive Physics Understanding Emerges from Self-Supervised Pretraining on Natural VideosQ. Garrido, N. Ballas, M. Assran, et al. · 2025 · arXiv
  AdvancedEvaluation
• TD-MPC2: Scalable, Robust World Models for Continuous ControlN. Hansen, H. Su, X. Wang · 2024 · ICLR
  AdvancedPlanning
• Diffusion Forcing: Next-token Prediction Meets Full-Sequence DiffusionB. Chen, D. M. Martínez, et al. · 2024 · NeurIPS
  AdvancedVideo
• GameNGen: Diffusion Models Are Real-Time Game EnginesD. Valevski, Y. Leviathan, M. Arar, S. Fruchter · 2024 · arXiv
  AdvancedVideo
• Dynamic 3D Gaussians: Tracking by Persistent Dynamic View SynthesisJ. Luiten, G. Kopanas, B. Leibe, D. Ramanan · 2024 · 3DV
  Advanced3D
• 4D Gaussian Splatting for Real-Time Dynamic Scene RenderingG. Wu, T. Yi, J. Fang, et al. · 2024 · CVPR
  Advanced3D
• OpenVLA: An Open-Source Vision-Language-Action ModelM. J. Kim, K. Pertsch, S. Karamcheti, et al. · 2024 · CoRL
  AdvancedRobotics
• Open X-Embodiment: Robotic Learning Datasets and RT-X ModelsOpen X-Embodiment Collaboration · 2024 · ICRA
  AdvancedRobotics
• Diffusion for World Modeling: Visual Details Matter in Atari (DIAMOND)E. Alonso, A. Jelley, V. Micheli, et al. · 2024 · NeurIPS
  AdvancedVideo
• DINO-WM: World Models on Pre-trained Visual Features enable Zero-shot PlanningG. Zhou, H. Pan, Y. LeCun, L. Pinto · 2024 · arXiv
  AdvancedPlanning
• Grounding Image Matching in 3D with MASt3RV. Leroy, Y. Cabon, J. Revaud · 2024 · ECCV
  Advanced3D
• Is Sora a World Simulator? A Comprehensive Survey on General World Models and BeyondZ. Zhu, X. Wang, W. Zhao, et al. · 2024 · arXiv
  AdvancedSurvey
• Evaluating the World Model Implicit in a Generative ModelK. Vafa, J. Y. Chen, A. Rambachan, et al. · 2024 · arXiv
  AdvancedTheory
• Transformers Use Causal World Models in Maze-Solving TasksA. F. Spies, W. Edwards, M. I. Ivanitskiy, et al. · 2024 · arXiv
  AdvancedTheory
• Video as the New Language for Real-World Decision MakingS. Yang, J. Walker, J. Parker-Holder, et al. · 2024 · arXiv
  AdvancedVideo
• Flow Matching for Generative ModelingY. Lipman, R. T. Q. Chen, H. Ben-Hamu, M. Nickel, M. Le · 2023 · ICLR
  AdvancedVideo
• GAIA-1: A Generative World Model for Autonomous DrivingA. Hu, L. Russell, H. Yeo, et al. · 2023 · arXiv
  AdvancedVideo
• Transformers are Sample-Efficient World Models (IRIS)V. Micheli, E. Alonso, F. Fleuret · 2023 · ICLR
  AdvancedDynamics
• Emergent Linear Representations in World Models of Self-Supervised Sequence ModelsN. Nanda, A. Lee, M. Wattenberg · 2023 · arXiv
  AdvancedTheory
• Language Models Represent Space and TimeW. Gurnee, M. Tegmark · 2023 · arXiv
  AdvancedTheory
• Learning Interactive Real-World Simulators (UniSim)S. Yang, Y. Du, K. Ghasemipour, et al. · 2023 · arXiv
  AdvancedVideo
• Video Diffusion ModelsJ. Ho, T. Salimans, A. Gritsenko, W. Chan, M. Norouzi, D. J. Fleet · 2022 · NeurIPS
  AdvancedVideo
• DayDreamer: World Models for Physical Robot LearningP. Wu, A. Escontrela, D. Hafner, K. Goldberg, P. Abbeel · 2022 · CoRL
  AdvancedRobotics
• R3M: A Universal Visual Representation for Robot ManipulationS. Nair, A. Rajeswaran, V. Kumar, C. Finn, A. Gupta · 2022 · CoRL
  AdvancedRobotics
• BEHAVIOR-1K: A Benchmark for Embodied AI with 1,000 Everyday Activities and Realistic SimulationC. Li, R. Zhang, J. Wong, et al. · 2022 · CoRL
  AdvancedEvaluation
• Emerging Properties in Self-Supervised Vision Transformers (DINO)M. Caron, H. Touvron, I. Misra, et al. · 2021 · ICCV
  AdvancedRepresentation
• Mastering Atari with Discrete World Models (DreamerV2)D. Hafner, T. Lillicrap, M. Norouzi, J. Ba · 2021 · ICLR
  AdvancedDynamics
• Object-Centric Learning with Slot AttentionF. Locatello, D. Weissenborn, T. Unterthiner, et al. · 2020 · NeurIPS
  AdvancedRepresentation
• Model-Based Reinforcement Learning for Atari (SimPLe)Ł. Kaiser, M. Babaeizadeh, P. Miłoś, et al. · 2020 · ICLR
  AdvancedDynamics
• When to Trust Your Model: Model-Based Policy Optimization (MBPO)M. Janner, J. Fu, M. Zhang, S. Levine · 2019 · NeurIPS
  AdvancedPlanning
• Representation Learning with Contrastive Predictive CodingA. van den Oord, Y. Li, O. Vinyals · 2018 · arXiv
  AdvancedRepresentation
• VINS-Mono: A Robust and Versatile Monocular Visual-Inertial State EstimatorT. Qin, P. Li, S. Shen · 2018 · IEEE T-RO
  AdvancedRobotics
• PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric SpaceC. R. Qi, L. Yi, H. Su, L. J. Guibas · 2017 · NeurIPS
  Advanced3D
• A Multi-State Constraint Kalman Filter for Vision-aided Inertial Navigation (MSCKF)A. I. Mourikis, S. I. Roumeliotis · 2007 · ICRA
  AdvancedRobotics