Home // DBKDA 2025, The Seventeenth International Conference on Advances in Databases, Knowledge, and Data Applications // View article

Route Planning in Wildfire Areas by Integrating a Modified A* Algorithm with Deep Learning

Authors:
Manavjit Singh Dhindsa
Kshirasagar Naik
Pin-Han Ho
Marzia Zaman
Chung-Horng Lung
Srinivas Sampalli

Keywords: Route Navigation; Deep Learning; Forest fire; A* Algorithm; Path Planning; Wildfire Prediction; Machine Learning.

Abstract:
Wildfires pose a significant threat to life, property, and ecosystems, with their frequency and intensity escalating due to climate change. Effective evacuation planning is critical to mitigating wildfire impacts, yet it remains a challenging task in dynamic, high-risk scenarios. This paper presents a framework for safe path planning that integrates wildfire spread predictions from state-of-the-art deep learning models with an optimized A* (OA*) algorithm. The proposed approach utilizes binary fire masks to generate safe and efficient evacuation routes while adhering to strict safety constraints, such as maintaining buffer zones around fire-affected regions. Experimental results show the algorithm’s capability to generate actionable paths and accurately identify no-path scenarios under diverse wildfire conditions. This framework offers a robust solution for real-time evacuation planning, contributing to the broader efforts of wildfire management and disaster mitigation.

Pages: 7 to 12

Copyright: Copyright (c) IARIA, 2025

Publication date: March 9, 2025

Published in: conference

ISSN: 2308-4332

ISBN: 978-1-68558-244-9

Location: Lisbon, Portugal

Dates: from March 9, 2025 to March 13, 2025