A GLS approach for Origin-Destination Matrices Estimation accounting for Spatio-Temporal Flexibility and Congestion

Abstract

This paper introduces an enhanced framework for Origin-Destination Matrices Estimation (ODME) using the Flex-GLS model, which integrates state-specific flexibility parameters to account for varying congestion levels. Building on prior research, the study leverages Floating Car Data (FCD) and Google Popular Times (GPT) to classify travel demand into macro-activities characterized by spatio-temporal flexibility metrics, which are further utilized to enhance the ODME process. The extended model incorporates network state types, predicted using Gaussian Processes, to dynamically adjust flexibility parameters according to prevailing traffic conditions. This methodology is validated through a case study in the EUR district of Rome, utilizing extensive FCD datasets spanning 2020 and 2023. Results demonstrate that Flex-GLS outperforms traditional GLS, offering more accurate demand estimation and link flow reproduction. Moreover, the study highlights the critical relationship between congestion levels and flexibility parameters, emphasizing the model’s adaptability to real-world urban mobility challenges.