SEED: A Single-Source Approach for the Endogenous OD-Estimation and Calibration of Travel Demand Models

• Applicants: Professor Dr.-Ing. Bernhard Friedrich; Professor Dr.-Ing. Markus Friedrich
• Subject Area: Traffic and Transport Systems, Intelligent and Automated Traffic ; City Planning, Spatial Planning, Transportation and Infrastructure Planning, Landscape Planning
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 551912017

Knowledge about current and future travel demand is important information for transport planning and the design of transport infrastructure. To obtain this information, travel demand models are used, such as the widely used four-stage model and more recent agent-based models. As exogenous input the models use information on transport supply, land use and mobility behavior. A major problem with traffic demand models is the inability to validate the resulting trips, as no data source is available to observe the actual trips. In practice, traffic modellers attempt to adjust both the size of demand and network capacity to replicate the observed traffic volumes at a small number of traffic counting points. Given the large number of possible solutions and the inconsistency of the observed values at different points in time, calibration and validation of the model in this way is only possible to a limited extent. To address this problem, various researches have looked at the so-called endogenous estimation of origin-destination relationships (OD matrix) in networks, which use available observed traffic volumes as a starting point instead of survey data. Consequently, endogenous models for the estimation of the OD matrix do not provide information on the causes of trips and cannot be used to determine the impact of future infrastructural or demographic developments. They are therefore only suitable for transport planning tasks to a limited extent. However, in combination with travel demand models, they offer a promising opportunity to improve their calibration and validation. As obvious and plausible as this combination of traditional travel demand modelling and endogenous models appears, its application currently fails due to the lack of availability and spatial-temporal consistency of traffic counts. In recent years, the emergence of mobile devices that can track the movements of vehicles and people has led to a massive increase in the availability of data. While this data is available almost everywhere and provides small but long-running samples at low cost, it does not contain information about traffic volumes and has its own noise and biases that need to be taken into account. However, recent research shows that this movement data provides insights into a wide range of traffic flow and network characteristics. In particular, the movement data can be used to derive the capacities of individual traffic facilities and thus provide essential constraints for modelling traffic demand. Given this background, the objective of the research project is to develop a methodology that uses travel data as a single and consistent source to derive a reliable OD matrix for a historical situation using an endogenous model. This demand estimate should then serve both as ‘ground truth’ and as additional information for exogenous travel demand models and should thus offer a first end-to-end solution for the calibration and validation of travel demand models.

• DFG Programme: Research Grants