SyncWorld: Visual Calibration Enables World Models as Zero-Shot Simulators

Hugging Face Daily Papers Papers
world-models visual-calibration zero-shot-simulation robotics action-conditioned policy-improvement simulators

Summary

SyncWorld is an action-conditioned world model that uses visual calibration episodes to learn environment-specific action-to-visual mappings, enabling zero-shot simulation and test-time policy improvement across unseen robotic settings.

World models are increasingly used as policy-in-the-loop imagination environments, where reliable rollouts require fine-grained controllability with respect to low-level robot actions. A key obstacle to scaling such models in robotics is that actions are not a universal language in pixel space: changes in visual environment, camera view, robot placement, or embodiment alter how the same numerical action manifests visually, leading to conflicting supervision under mixed training and brittle generalization at deployment. We introduce SyncWorld, an action-conditioned world model that serves as a zero-shot simulator across unseen environments without any additional training. SyncWorld leverages a visual calibration episode---paired frames and actions that showcase all the controllable degrees of freedom---to specify the setup-specific Action--Visual Mapping in context. Training with visual calibration contexts teaches the model to interpret actions through visual evidence and to leverage interaction history when explicit calibration is unavailable. Experiments show that SyncWorld can accurately simulate action outcomes in previously unseen settings, and that its capability of simulating rollouts enables test-time policy improvement without training.
Original Article
View Cached Full Text

Cached at: 09/10/26, 10:11 AM

Paper page - SyncWorld: Visual Calibration Enables World Models as Zero-Shot Simulators

Source: https://huggingface.co/papers/2609.09155

Abstract

SyncWorld is an action-conditioned world model that uses visual calibration episodes to learn environment-specific action-to-visual mappings, enabling zero-shot simulation and test-time policy improvement across unseen robotic settings.

World modelsare increasingly used as policy-in-the-loop imagination environments, where reliable rollouts require fine-grained controllability with respect to low-level robot actions. A key obstacle to scaling such models in robotics is that actions are not a universal language in pixel space: changes in visual environment, camera view, robot placement, or embodiment alter how the same numerical action manifests visually, leading to conflicting supervision under mixed training and brittle generalization at deployment. We introduce SyncWorld, anaction-conditionedworld model that serves as azero-shot simulatoracross unseen environments without any additional training. SyncWorld leverages avisual calibrationepisode---paired frames and actions that showcase all the controllable degrees of freedom---to specify the setup-specificAction--Visual Mappingin context. Training withvisual calibrationcontexts teaches the model to interpret actions through visual evidence and to leverage interaction history when explicit calibration is unavailable. Experiments show that SyncWorld can accurately simulate action outcomes in previously unseen settings, and that its capability of simulating rollouts enables test-timepolicy improvementwithout training.

View arXiv pageView PDFProject pageGitHub7Add to collection

Get this paper in your agent:

hf papers read 2609\.09155

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper1

#### yyuncong/SyncWorld Image-to-Video• 16B• Updated1 day ago • 51 • 1

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2609.09155 in a dataset README.md to link it from this page.

Spaces citing this paper1

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

World in World: Explore the World with World Models

Hugging Face Daily Papers

The paper presents World in World, a training-free interface that enables flexible camera and time control in frozen autoregressive video world models by using correspondence-guided queries and evidence-wise attention guidance.

WorldCycle: Self-Verifiable Reinforcement Learning for Long-Horizon Video World Models

Hugging Face Daily Papers

WorldCycle proposes a self-verifiable reinforcement learning method for long-horizon video world models, using reversible action cycles as free supervision to reduce state-returning drift by up to 44% and boost composite-action accuracy nearly 4x. It also introduces CycleBench to evaluate world models as simulators.