Optimizing the design of a hybrid urban mobility system

Aim of the project

The project aimed to develop an integrated optimization framework for designing efficient and accessible hybrid urban mobility systems by combining conventional fixed-route public transport (FRT) with flexible demand-responsive transit (DRT) services. Traditionally, these services are planned independently, overlooking important interdependencies between network design, passenger demand, and operational costs. The project addressed this challenge by developing optimization methods that jointly considered strategic public transport network design and operational planning of demand-responsive services. The overall objective was to improve urban accessibility and service quality while minimizing the resources required to operate the integrated transport system.

The project was conducted in collaboration between Vrije Universiteit Brussel (VUB), through the Mobilise Research Group, and Hasselt University (UHasselt), through the Logistics Research Group (LOG), with funding from the Research Foundation – Flanders (FWO).

Results of the project

The project developed an integrated optimization framework for the joint design of fixed-route and demand-responsive transit systems. A comprehensive literature review identified key gaps in existing integration approaches, particularly the limited consideration of joint strategic and operational planning. Building on these findings, a bi-level optimization model was developed to jointly optimize fixed-route network configurations and demand-responsive vehicle operations. Computational experiments demonstrated that integrated planning could reduce total fleet requirements by approximately 11% compared to conventional sequential approaches, while maintaining comparable service coverage and passenger travel times. The project further introduced machine learning–enhanced optimization methods to improve computational efficiency, achieving substantial reductions in runtime while preserving solution quality.

The Mobilise Research Group at VUB, in collaboration with the Logistics Research Group at UHasselt, developed integrated optimization and machine learning approaches for the joint design of fixed-route and demand-responsive transit systems. The research focused on improving network design, operational efficiency, and resource allocation under different travel demand patterns, providing scalable decision-support methods for more efficient and accessible public transportation.

The project developed a bi-level optimization framework combining Particle Swarm Optimization (PSO) for fixed-route network design with Large Neighborhood Search (LNS) for demand-responsive vehicle routing and scheduling.

A machine learning–enhanced optimization framework was also developed, using surrogate models, particularly XGBoost, to accelerate computationally expensive operational evaluations. A hybrid approach combining machine learning predictions with periodic explicit optimization achieved a 4–6 times reduction in computational time, while maintaining high solution quality.

These frameworks provide a methodological foundation for future decision-support tools for integrated public transportation planning.

Publications

Project team members

Dr. Seyedehsima Madani – VUB/UHasselt, PhD researcher

Prof. Imre Keserü – VUB, PhD supervisor

Prof. Kris Braekers – UHasselt, PhD supervisor