The Flow has not summarised this story yet — read it at arXiv Machine Learning.
The paper introduces a simulation‑grounded vision‑language model (VLM) framework for wildfire monitoring that converts 2D wildfire simulations into labeled video episodes using a fixed Blender mapping to create low‑detail 3D proxies. These proxies, along with controllable video generation, provide a multimodal memory that a training‑free multi‑agent VLM system uses to retrieve reference episodes, reconcile visual and memory‑based predictions, and generate structured wildfire reports. The system achieves 77.3% accuracy on six simulator‑derived report fields, outperforming direct VLM querying and text‑only memory baselines.
arXiv:2608. 00012v1 Announce Type: cross Abstract: Multimodal Large Language Models (MLLMs) are increasingly used to interpret Earth observation data, yet their capability to support real-world disaster emergency response remains insufficiently evaluated.
The paper presents modular deep learning augmentations for next‑day wildfire spread prediction, including wind‑ and slope‑conditioned attention biases, physics‑feature retrieval‑augmented output correction, and fire‑conditioned dual‑stream gating. These modules are evaluated on five backbone models using the Next Day Wildfire Spread benchmark, with staged ablations, directional audits, retrieval perturbations, calibration measures, and computational comparisons. The best augmented SwinUNETR model achieves an F1 score of 0.4216 and an AUC‑PR of 0.3673, while a mixed ensemble reaches 0.4292 and 0.3790, demonstrating that predictive performance, operational trustworthiness, and computational practicality can be simultaneously improved.
arXiv:2607. 21444v1 Announce Type: cross Abstract: Reliable electric vehicle (EV) charging infrastructure is a cornerstone of sustainable, low-carbon cities, yet urban climate stress such as extreme heat, heavy precipitation, and humidity increasingly raises equipment fault risk and undermines the resilience of urban energy and mobility services.
arXiv:2609.23064v1 Announce Type: new Abstract: Understanding the physical world requires more than object recognition, scene description, and short-term visual prediction, as real-world physical sys...
The paper presents a modular, scalable framework for estimating the probability of failure (PoF) of power‑line assets using geospatial machine learning. It integrates diverse environmental predictors—topography, vegetation indices, lightning climatology, proximity features, and operational records—to model vegetation‑ and lightning‑related failure modes. The architecture is designed to be computationally efficient, easily extensible to new data sources, and suitable for large‑scale utility deployment, enabling asset‑level risk stratification for inspection and resilience planning.