On-Demand Multimodal Transit Systems

header-_bannerfinal.pngOn-Demand Multimodal Transit

In the United States, car ownership remains the best predictor of upwards social mobility. Those without a car are grievously disadvantaged in accessing jobs, health care, and decent groceries. Moreover, housing patterns further limit social mobility, as low-income populations often reside far away from job opportunities and have few efficient public transit options. Ride-hailing services have sometimes helped in providing additional mobility options. But, in general, they have widened inequalities in accessibility, servicing the needs of an affluent population, reducing the revenues of transit authorities, and increasing congestion and greenhouse gas emissions.

This project about On-Demand Multimodal Transit Systems (ODMTS) envisions a potential future for public transit that meets these mobility challenges for all population segments, by jointly addressing accessibility, convenience, affordability, congestion, and environmental issues. ODMTS combine on-demand services to serve low-density regions with high-occupancy vehicles (buses and/or trains) to travel along high-density corridors. The resulting door-to-door services differ from micro-mobility solutions by planning, operating, and optimizing transit systems holistically, using state-of-the-art optimization technology and machine learning.

Here is how they work!

ODMTS have been shown to improve convenience (e.g., transit times) and reduce costs for a variety of transit systems, including for the cities of Canberra in Australia and the broader Ann Arbor and Ypsilanti region. They provide a unique opportunity to expand services and improve job accessibility in neighborhoods where traditional transit systems have been too costly. The project now focuses on the Atlanta metropolitan area, which raises fundamental challenges in mobility. In the past year, the project demonstrated, through high-fidelity simulations, that ODMTS can bring substantial benefits in convenience, cost, and accessibility to all population segments. The ability of ODMTS to serve low-density areas effectively and to leverage economy of scale, as well as their agility, make them particularly suited for a post COVID-19 world. These results will be described shortly on this page!

wshuttle.png

Shuttle

Shuttles act as a first and last mile solution offering door-to-door services

wbus.png

Bus

Bus rapid transit connects users to rail stations, complementing rail

wbus.png

Rail

The backbone of the system, which runs on a fixed schedule          

Impact

Convenience

impact-_equity.png

The system is door-to-door, while an app guides them through the trip: it typically reduces transit time substantially.

Cost

impact-_equity.png

The system reduces cost through its multimodal design, leveraging rail and rapid bus transit on high-density corridors and small vehicles to serve the first and last miles.

Equity

impact-_equity.png

By reducing cost and improving convenience, ODMTS should make transit available, reliable, and fast for all neighborhoods. 

Simulation

 

This animation shows our preliminary design for an ODMTS in Atlanta: It describes the operations of the shuttles, buses, and rail between 7:00 am and 10:00am using the existing ridership. Stay tunes for more visualizations in the next months.

 

Publications

 

In Press

Spatio-Temporal Point Processes with Attention for Traffic Congestion Event Modeling. Shixiang Zhu, Ruyi Ding, Minghe Zhang, Pascal Van Hentenryck, and Yao Xie.  Transactions on Intelligent Transportation Systems.


2021

Branch and Price for Bus Driver Scheduling with Complex Break Constraints. Lucas Kletzander, Nysret Musliu, and Pascal Van Hentenryck. Proceeding of The Thirty-Fifth AAAI Conference on Artificial Intelligence (AAAI-21), February, 2021.


Commuting with Autonomous Vehicles: A Branch and Cut Algorithm with Redundant Modeling, Mohd. Hafiz Hasan and Pascal Van Hentenryck. ArXiv:2101.01072.


Capturing Travel Mode Adoption in Designing On-demand Multimodal Transit Systems. Beste Basciftci and Pascal Van Hentenryck. ArXiv:2101:01056.


2020

Real-Time Dispatching of Large-Scale Ride-Sharing Systems: Integrating Optimization, Machine Learning, and Model Predictive Control. Connor Riley, Pascal Van Hentenryck, and Enpeng Yuan. In the Proceedings of 29th International Joint Conference on Artificial Intelligence (IJCAI-20), Tokyo, Japan 2020.


Bilevel Optimization for On-Demand Multimodal Transit Systems. Beste Basciftci and Pascal Van Hentenryck. In the Proceedings of the 17th International Conference on the Integration of Constraint Programming, Artificial Intelligence, and Operations Research (CPAIOR 2020), Vienna, Austria, May, 2020.


Transfer-Expanded Graphs for On-Demand Multimodal Transit Systems. Kevin Dalmeijer and Pascal Van Hentenryck. In the Proceedings of the 17th International Conference on the Integration of Constraint Programming, Artificial Intelligence, and Operations Research (CPAIOR 2020), Vienna, Austria, May, 2020. 


Shared E-scooters: Business, Pleasure, or Transit? William Espinoza, Matthew Howard, Julia Lane, Pascal Van Hentenryck,  Transportation Board Annual Meeting, Washington DC, January 2020.


Optimization Models for Estimating Transit Network Origin-Destination Flows with AVL/APC Data. Xinyu Liu, Pascal Van Hentenryck, and Xilei Zhao. Transportation Board Annual Meeting, January 2020.


Distilling Black-Box Travel Mode Choice Model for Behavioral Interpretation. Xilei Zhao, Zhengze Zhou, Xiang Yan, Pascal Van Hentenryck. In the 2020 Transportation Board Annual Meeting, January 2020.


Joint Vehicle and Crew Routing and Scheduling. Edward Lam, Pascal Van Hentenryck, and Phil Kilby, Transportation Science, 54(2), Pages 299-564. March-April, 2020.


Prediction and Behavioral Analysis of Travel Mode Choice: A Comparison of Machine Learning and Logit Models. Xiang Yan, Xilei Zhao, Alan Yu, Pascal Van Hentenryck.  Travel Behaviour and Society. Volume 20, Pages 22-35, July 2020.


2019

On-Demand Mobility Systems, Pascal Van Hentenryck, ISE Magazine, Fall 2019.


The Flexible and Real-Time Commute Trip Sharing Problems. Hafiz Hasan and Pascal Van Hentenryck. In the 24th International Conference on the Principles and Practice of Constraint Programming, Stamford, CT, September 2019.


A Column Generation for Online Ride-Sharing Services. Connor Riley, Antoine Legrain, and Pascal Van Hentenryck. In the Proceedings of 16th International Conference on the Integration of Constraint Programming, Artificial Intelligence, and Operations Research (CPAIOR 2019), Thessaloniki, Greece, June 4–7, 2019.


Benders Decomposition for the Design of a Hub and Shuttle Public Transit System.  Arthur Maheo, Philip Kilby, and Pascal Van Hentenryck.  Transportation Science.  53(1), 77-88, January-February 2019.


2018

Constraint and Mathematical Programming Models for Integrated Port Container Terminal Operations, Damla Kizilay, Deniz T. Eliiyi, and Pascal Van Hentenryck. 15th International Conference on the Integration of Constraint Programming, Artificial Intelligence, and Operations Research, Delft, The Netherlands, June 26-29, 2018.


2016

A Branch-and-Price-and-Check Model for the Vehicle Routing Problem with Location Resource constraints. Edward Lam and Pascal Van Hentenryck.  Constraints 21(3): 394-412 (2016) (Fast-Track Paper from CPAIOR’06).


Optimizing Infrastructure Enhancements for Evacuation Planning. Kunal Kumar, Julia Romanski, and Pascal Van Hentenryck. Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence (AAAI-16), Phoenix, Arizona, February, 2016.


A Conflict-Based Path-Generation Heuristic for Evacuation Planning. Victor Pillac, Pascal Van Hentenryck, and Caroline Even. Transportation Research Part B., 83, 136-150, January, 2016.


2015

A Constraint Programming Approach for Non-Preemptive Evacuation Scheduling Caroline Even, Andreas Schutt, and Pascal Van Hentenryck. Proceedings of the International Conference on Principles and Practice of Constraint Programming, Cork, Ireland.


A Column-Generation Approach for Joint Mobilization and Evacuation Planning Victor Pillac, Manuel Cebrian, and Pascal Van Hentenryck. Constraints, 20(3), 285-303, July 2015. (fast-track paper from CPAIOR’15).


A Multi-Stage Very Large-Scale Neighborhood Search for the Vehicle Routing Problem with Soft Time-Windows Sebastien Mouthuy, Florence Massen, Yves Deville, Pascal Van Hentenryck. Transportation science, 49(2), 223-238, May 2015


2014

Multi-period vehicle loading with stochastic release dates Tasemin Arda, Yves Crama, David Kronus, Thierry Pironet, Pascal Van Hentenryck. EURO Journal on Transportation and Logistics, pp. 93119, 10.1007/s13676-013-0035-z.


2012

Randomized Adaptive Vehicle Decomposition for Large-Scale Power Restoration Ben Simon, Carleton Coffrin, Pascal Van Hentenryck. Proceedings of the International Conference on Integration of Artificial Intelligence and Operations Research Techniques in Constraint Programming for Combinatorial Optimization Problems (CPAIOR), Nantes, France, pp. 379-394, May, 2012.


Collaborators

marta logo
city-of-atlanta logo
arc logo
gdot logo
tollway logo
midtown logo