Research Seminars 2020/2021

Maryam DARVISH - Université Laval, Canada

March 30th 2021 - Online

Logic-Based Benders algorithms for a Time-Depenent Vehicle Routing Problem

Abstract: The classic Vehicle Routing Problem (VRP) is very well studied in the operations research literature. Time-Dependent Vehicle Routing Problem (TDVRP) is an extension of the VRP, which has gained considerable attention during the past years due to its applicability to transportation planning in urban areas. In this talk, we present two models for the TDVRP and show the results obtained by a commercial solver for instances generated based on Quebec City’s road network. Furthermore, we propose a Logic-Based Benders decomposition framework and demonstrate its success in obtaining quality solutions.

Leandro C. COELHO - Université Laval, Canada

March 4th 2021 - Online

Traffic and the city: Traffic, Data and Optimization

Abstract: All major cities face problems related to the number of cars on the streets, urban congestion, and lost hours in traffic. This leads to longer travel times, uncertainties about the best routes, irregular public transportation, among others. In this presentation we will see how massive data can help us correct these and other problems, what techniques and operations have been used, and what trends and opportunities for the future. We describe a practical approach to measure and compute not only traffic but also the time it takes to traverse any arc in a large network based on a large amount of data obtained through crowdsourcing. Our method is capable of providing the state of the network with 15-minute intervals for any day of the week. Using this large database, we propose new algorithms for time-dependent quickest path (point to point) and for vehicle routing problems with a comprehensive objective function encompassing not only distance, but also time, costs, and emissions. We also compare the results of classical emission estimation models against new machine learning tools that can significantly better predict emissions incurred by driving in different conditions.