A landmark-based addressing framework for urban navigation using geospatial clustering and pathfinding algorithm

Abstract

Urban navigation in rapidly growing cities often faces challenges due to incomplete addressing systems, especially in cities like Kathmandu, Nepal, where traditional street-based systems are unreliable. This study proposes a landmark-based addressing framework that integrates culturally significant landmarks with modern geospatial tools such as OpenStreetMap (OSM), GeoPandas, Hierarchical Hexagonal Indexing (H3), Density-Based Spatial Clustering of Applications with Noise (DBSCAN), and A* search for optimized pathfinding, supported by PostgreSQL and its spatial extension, PostGIS for scalable data management. A web-based interface built with Leaflet.js and FastAPI provides real-time access to landmark-based navigation tools. Simulation results, conducted on a comprehensive dataset of 149,054 buildings in Kathmandu, reveal that the landmark-based system significantly outperforms traditional approaches. The average path length was reduced by 37.7% (from 69.22 to 43.12 nodes), and the average travel time decreased by 22.9% (from 550.86 to 424.92 seconds). This system offers a practical and scalable solution for urban navigation, emergency response, and service delivery in cities with informal or incomplete addressing infrastructures.