CF-JEPA is a JEPA-style latent world model that separates the latent space into controllable and uncontrollable subspaces, allowing distractor information to be captured in the uncontrollable region. This factorization prevents latent collapse and maintains performance across 2D and 3D control tasks, even under distracted conditions. The model is validated on a simulated robot task, demonstrating its practical applicability.
arXiv:2608. 06544v1 Announce Type: new Abstract: World models for visual control typically learn compact latent states by reconstructing observations, implicitly encouraging representations to preserve information across the entire visual input.
Contrastive World Models propose a new method for learning latent dynamics without pixel reconstruction. By replacing observation reconstruction with a Deep InfoMax-like objective that maximizes mutual information between state-action sequences and local patch features of future observations, the approach encourages state representations to retain predictive information while ignoring visually irrelevant details. Experiments show that this method matches existing baselines in simple settings and significantly outperforms them when distractors or natural video backgrounds are present, while also training more efficiently by eliminating the pixel decoder.
JEPA-Bisim introduces a bisimulation encoder to joint-embedding predictive world models, ensuring that states with similar transition dynamics are mapped to nearby latent representations while suppressing irrelevant slow features such as background changes and distractors. The approach improves robustness on navigation (PointMaze) and manipulation (PushT) tasks under varied test-time visual conditions, achieving up to tenfold smaller latent spaces than DINO-WM. It remains effective across different pre-trained visual encoders, including DINOv2, SimDINOv2, and iBOT.
arXiv:2608.29434v1 Announce Type: cross Abstract: JEPA world models make latent-space planning a practical route to control, but they are built almost exclusively on images. Whether latent prediction...
arXiv:2608. 06706v1 Announce Type: cross Abstract: Latent world models plan by predicting future states from an action, but when a scene contains motion the agent does not control, they quietly go action-blind: predictions for different actions become indistinguishable even as the training loss keeps improving.
arXiv:2606. 12217v1 Announce Type: cross Abstract: World Action Models (WAMs) offer a promising route for robot manipulation by using video generation models to model future scene evolution before producing control actions.
arXiv:2606. 15768v1 Announce Type: cross Abstract: Vision-Language-Action models (VLAs) leverage large-scale vision-language pretraining for semantic robot control, but often lack explicit foresight into how robot actions change the scene.
StarWM introduces a self‑supervised attention routing mechanism that selectively applies reconstruction only to dynamically relevant regions of visual input. By combining a cross‑attention module with a dual‑stream decoder and stop‑gradient barriers, it balances faithful environmental dynamics capture with abstraction of irrelevant content. Experiments on DeepMind Control show that StarWM outperforms both reconstruction‑based and reconstruction‑free baselines, especially under distractor conditions, and preserves state attributes over long‑horizon imagination.
The paper introduces VIGOR, a framework for zero‑shot visual generalization in model‑based reinforcement learning. VIGOR enforces latent‑space consistency through asymmetric weak‑to‑strong augmentations, dynamics‑level consistency, and encoder‑level stabilization, allowing the agent to handle unseen visual distractions while maintaining sample efficiency. Experiments on the DeepMind Control Suite and Robosuite demonstrate that VIGOR outperforms state‑of‑the‑art baselines, achieving significant gains in both environments.
arXiv:2605.15618v2 Announce Type: replace-cross Abstract: Self-supervised video models are increasingly framed as world models, yet they are still evaluated almost entirely on clean video and reporte...
arXiv:2609.16864v1 Announce Type: cross Abstract: Vision-language-action (VLA) models have achieved impressive performance in quasi-static manipulation, but struggle in dynamic manipulation tasks bec...
World Action Models (WAMs) offer a promising route for robot manipulation by using video generation models to model future scene evolution before producing control actions. However, our empirical observations reveal a phenomenon: generating plausible visual futures does not always guarantee the extraction of accurate actions.