Title: Optimizing Urban Traffic Simulations with Sparse Data Using Genetic Algorithms

Urban traffic simulation plays a pivotal role in infrastructure planning, such as determining the optimal locations for electric vehicle charging stations. However, creating accurate simulations for numerous cities is often impeded by significant data constraints: comprehensive traffic measurements exist for only a limited number of road segments in most municipalities, and employment data necessary for modeling commuter flows is seldom available at the fine resolution required for simulation.

To overcome these hurdles, this study introduces a framework grounded in genetic algorithms that calibrates urban traffic models using sparse road observations, eliminating the need for granular job location data. By utilizing the SUMO traffic simulation platform for Greensboro, North Carolina, the proposed method adjusts job distribution patterns and gate-traffic parameters to synchronize simulated traffic with traffic-flow rates from a limited set of monitored roads.

The results indicate that this technique yields simulated traffic that aligns closely with actual measurements. Furthermore, the model effectively generalizes to road segments excluded from the training set. Notably, the derived job distributions exhibit strong qualitative consistency with census employment data, even though the model was not explicitly trained on that specific information. This research highlights that realistic urban traffic simulation is feasible with minimal real-world data, presenting a scalable, data-efficient calibration method that lowers the barriers to implementing traffic models in a variety of urban environments.

Source: arXiv Generated at: 2026-06-03 00:00:00 UTC