PhysFire-WM: A Physics-Informed World Model for Emulating Fire Spread Dynamics

TL;DR

PhysFire-WM is a physics-informed world model that improves fire spread prediction by integrating combustion dynamics and thermal information, addressing limitations of previous binary mask models.

Contribution

The paper introduces PhysFire-WM, a novel physics-informed world model that incorporates structured priors from a physical simulator and a cross-task training strategy for accurate fire forecasting.

Findings

Outperforms existing models in fire spread accuracy

Enhances physical realism and geometric precision

Validates effectiveness on a multimodal fire dataset

Abstract

Fine-grained fire prediction plays a crucial role in emergency response. Infrared images and fire masks provide complementary thermal and boundary information, yet current methods are predominantly limited to binary mask modeling with inherent signal sparsity, failing to capture the complex dynamics of fire. While world models show promise in video generation, their physical inconsistencies pose significant challenges for fire forecasting. This paper introduces PhysFire-WM, a Physics-informed World Model for emulating Fire spread dynamics. Our approach internalizes combustion dynamics by encoding structured priors from a Physical Simulator to rectify physical discrepancies, coupled with a Cross-task Collaborative Training strategy (CC-Train) that alleviates the issue of limited information in mask-based modeling. Through parameter sharing and gradient coordination, CC-Train effectively integrates thermal radiation dynamics and spatial boundary delineation, enhancing both physical realism and geometric accuracy. Extensive experiments on a fine-grained multimodal fire dataset demonstrate the superior accuracy of PhysFire-WM in fire spread prediction. Validation underscores the importance of physical priors and cross-task collaboration, providing new insights for applying physics-informed world models to disaster prediction.