Language-Driven Semantic Change Detection in Urban Maps via Multi-Modal Deep Learning

Huaze Liu, Zihao Gao, Adyasha Mohanty

Abstract:	High-integrity maps are essential for safe autonomous navigation in dynamic urban environments, where frequent changes and sensor limitations present significant challenges. This paper introduces a novel, deep-learning-driven framework for continuous map uncertainty monitoring and semantic change detection. Our approach leverages data-driven feature extraction from both vision and LiDAR modalities. A key innovation is the integration of zero-shot semantic segmentation using large pre-trained vision-language models, which provides interpretable, language-driven explanations for detected map inconsistencies. The framework dynamically tracks map consistency using Kullback-Leibler divergence metrics, enabling proactive real-time alerts when deviations occur. By jointly assessing structural and semantic integrity, our approach provides a robust and interpretable mechanism for maintaining high-integrity maps in urban autonomous systems.
Published in:	Proceedings of the 38th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2025) September 8 - 12, 2025 Hilton Baltimore Inner Harbor Baltimore, Maryland
Pages:	1004 - 1023
Cite this article:	Liu, Huaze, Gao, Zihao, Mohanty, Adyasha, "Language-Driven Semantic Change Detection in Urban Maps via Multi-Modal Deep Learning," Proceedings of the 38th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2025), Baltimore, Maryland, September 2025, pp. 1004-1023. https://doi.org/10.33012/2025.20234
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In