Shortest Path Calibration: A Framework for Calibrating Open-Source Networks with Routing Services

This paper presents a novel approach to network modelling, focusing on the optimisation of routing solutions. The framework leverages OpenStreetMap (OSM) data for the automatic calibration of network models, with paths from routing services serving as a benchmark for optimization. The study introduces “routing factors” that adjust travel costs across different types of roads, aligning routes derived from OSM more closely with those suggested by the benchmark paths. The methodology involves extracting network data from OSM and transforming it into a graph for the computation of shortest paths. The calibration process employs the Simulated Annealing (SA) algorithm and the Root Mean Square Error (RMSE) metric to evaluate the similarity between OSM-generated routes and those obtained from the benchmark. A case study with a network of Rome demonstrates a significant reduction in RMSE following calibration, highlighting the effectiveness of this framework. In addition, the paper also validates the model for new random points. The significant contribution of this study to the field is the calibration of the supply model derived from OSM network data. This approach is particularly beneficial for small logistic companies, as it offers a data-driven, cost-effective solution for the automatic calibration of network models, enhancing routing efficiency and accuracy. This could potentially lead to significant cost savings for these companies.