Study on an Automatic Correction Mechanism for Urban Infrastructure Location Information Based on Operational and Movement Data

In the operation of digital twins for realizing smart cities, the inability of urban infrastructure location data to autonomously adapt to real-world changes hinders high-precision simulation and optimization. To address this challenge, we are investigating a method that automatically estimates accurate locations using indirect information—operation and activity data obtained from vehicles in service—rather than directly measuring the geographic information of urban infrastructure. This research enables stable position estimation by eliminating the effects of noise and outliers present in the data. Specifically, we verify a set of methods that first achieve automatic noise filtering and cluster separation using DBSCAN, then determine robust true object coordinates resistant to outliers using spatial medians. This paper applies the proposed methods to our garbage collection visualization system under development, quantitatively demonstrating the detection accuracy of discrepancies between master data and the latest real-world position information, as well as the robustness and effectiveness of each method. This demonstrates the feasibility of implementing data assimilation technology to maintain the accuracy and reliability of the digital twin platform.