Multi-objective Optimization of Intelligent Traffic Control, part of the Conductor project

Abstract

In recent years there has been growing interest in dynamic transport demand management and optimization of traffic and fleet management using dynamic models and algorithms. One of the biggest challenges for transport authorities is finding solutions that efficiently distribute network capacity over different transport modes, e.g. passenger cars, trucks, public transport, shared vehicles, bikes, and pedestrians, without seriously hampering traffic flow, travel preferences, and safety. Given priority to a particular mode should not be at the cost of degrading the level of service or safety for other road users in the short or long term. Meanwhile, we must consider individuals’ well-being and sustainable development goals (SGDs) when intervening in transport systems. This results in a multi-objective problem (e.g., travel time, delays, waiting time, fuel consumption, CO2 emission, and safety) and multi-stakeholder (e.g., citizens, logistic companies, and public authorities) that needs to be solved at different levels.

This Ph.D. project aims to develop dynamic multi-objective optimization models for multi-modal networks using Machine Learning (ML) that could improve network efficiency and users’ well-being.

The research involves optimizing multi-modal signal control with Freight Signal Prioritization (FSP) and conducting impact analyses on other road users. Furthermore, a pilot test will be conducted in Almelo, the Netherlands, where 27 intelligent traffic control systems (iVRIs) have been implemented that can effectively communicate with vehicles and road users. In the next step, the models will be upgraded to a network-wide multi-modal network optimization considering fixed-schedule modes (buses and trains), passenger cars, trucks, bikes, and shared modes. Furthermore, in a futuristic scenario, cooperative and autonomous vehicles will be integrated into the model in the simulation environment.