OpenPaths AGENT Implementation in Milan
Building a modern, transparent, and automated modelling platform with OpenPaths AGENT
citiEU was appointed by Agenzia Mobilità Ambiente Territorio (AMAT) to implement different features from OpenPaths AGENT, within their OpenPaths CUBE model. This included moving the Demand Model processes to OpenPaths AGENT, incorporating the AGENT autocalibration feature to use the results of Matrix Estimation, and adjusting the initial parameters of Generation, Distribution and Mode Choice simultaneously.
citiEU played a key technical role in the modelling by implementing the algorithms requested by AMAT and ensuring methodological consistency, managing calibration workflows and supporting the continuous updating and refinement of the model. The project was delivered with the scientific supervision of Professor Armando Cartenì from Università degli Studi della Campania.
The project aimed to tackle several persistent challenges:
​
-
Long calibration and scenario execution times, which historically obstructed rapid analysis and iteration.
-
Dependence on manual adjustments, and dependence on the quality of input data.
​
​
Our main goals were to build a modern, transparent, and automated modelling platform, reducing development and calibration overhead through automation and clean data structures.
The adoption of OpenPaths CUBE with AGENT significantly improved modelling transparency and collaboration, and the new automated calibration process performed reliably. Removing legacy adjustment coefficients did not harm model accuracy and increased flexibility in scenario design. Furthermore, GitHub-based collaborative editing and SQL integration reduced errors and enhanced team productivity.
​
A core application of the model is to support sustainable mobility planning by:
-
Quantifying impacts of active mobility policies, public transport improvements, and park-and-ride facilities.
-
Enabling scenario simulations that estimate modal shifts and their effects on COâ‚‚ emissions and vehicle-km reductions, and providing data-driven support for climate strategies at municipal and metropolitan levels.
Although not directly quantified in this project phase, the model creates a foundation to measure and monitor environmental KPIs (e.g., emissions, mode share changes) across scenarios.
​
The modular structure and integration with Python workflows (CubePy) allows for future enhancements such as real-time data integration, dashboard creation, and machine learning-based pattern detection. Furthermore, the model architecture is scalable, and lessons learned here can be reused in broader regional or national transport studies.
Contact Us
Get in touch so we can start working together.
