VDAWorld: World Modelling via VLM-Directed Abstraction and Simulation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: O'Mahony, Felix, Cipolla, Roberto, Tewari, Ayush
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908707050225664
author O'Mahony, Felix
Cipolla, Roberto
Tewari, Ayush
author_facet O'Mahony, Felix
Cipolla, Roberto
Tewari, Ayush
contents Generative video models, a leading approach to world modeling, face fundamental limitations. They often violate physical and logical rules, lack interactivity, and operate as opaque black boxes ill-suited for building structured, queryable worlds. To overcome these challenges, we propose a new paradigm focused on distilling an image caption pair into a tractable, abstract representation optimized for simulation. We introduce VDAWorld, a framework where a Vision-Language Model (VLM) acts as an intelligent agent to orchestrate this process. The VLM autonomously constructs a grounded (2D or 3D) scene representation by selecting from a suite of vision tools, and accordingly chooses a compatible physics simulator (e.g., rigid body, fluid) to act upon it. VDAWorld can then infer latent dynamics from the static scene to predict plausible future states. Our experiments show that this combination of intelligent abstraction and adaptive simulation results in a versatile world model capable of producing high quality simulations across a wide range of dynamic scenarios.
format Preprint
id arxiv_https___arxiv_org_abs_2512_11061
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle VDAWorld: World Modelling via VLM-Directed Abstraction and Simulation
O'Mahony, Felix
Cipolla, Roberto
Tewari, Ayush
Computer Vision and Pattern Recognition
Generative video models, a leading approach to world modeling, face fundamental limitations. They often violate physical and logical rules, lack interactivity, and operate as opaque black boxes ill-suited for building structured, queryable worlds. To overcome these challenges, we propose a new paradigm focused on distilling an image caption pair into a tractable, abstract representation optimized for simulation. We introduce VDAWorld, a framework where a Vision-Language Model (VLM) acts as an intelligent agent to orchestrate this process. The VLM autonomously constructs a grounded (2D or 3D) scene representation by selecting from a suite of vision tools, and accordingly chooses a compatible physics simulator (e.g., rigid body, fluid) to act upon it. VDAWorld can then infer latent dynamics from the static scene to predict plausible future states. Our experiments show that this combination of intelligent abstraction and adaptive simulation results in a versatile world model capable of producing high quality simulations across a wide range of dynamic scenarios.
title VDAWorld: World Modelling via VLM-Directed Abstraction and Simulation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.11061