Route choice behavior in Mobility-as-a-Service (MaaS) systems are governed by very different factors than conventional transportation systems; instead of the link cost being dependent on flow, it is the capacity. This study with Susan Jia Xu introduces “congestible capacities” in route choice and shows how we can model this phenomenon using flow-capacity input-output systems of equations and estimate their parameters in an offline-online estimation method. Data from Citi Bike are used to demonstrate this approach and show that online route choice models that account for congestible capacities are more accurate and can explain the observed changes in MaaS systems.

The study was funded by NSF CMMI-1634973 and CMMI-1652735.
https://ieeexplore.ieee.org/document/9520135

The team developed a nonmyopic online algorithm for rebalancing electric carshare services with explicit charging capacities (which can be readily adapted to other on-demand mobility operations) in this international collaboration with Luxembourg Institute of Socio-Economic Research and NYU Abu Dhabi. Simulation experiments using demand data shared by BMW ReachNow Brooklyn show that the proposed algorithm reduces the cost increase of switching to electric vehicles (EV) by 38% compared to myopic algorithms. This might make it more viable for fleets (both passenger and freight) to consider switching to EV, which is especially vital for environmental sustainability in this time of climate change as the U.S. moves toward more EV infrastructure investments.

The research was funded by C2SMART (USDOT award #69A3551747124, Luxembourg National Research Fund [Grant INTER/MOBILITY/17/11588252], and the NYUAD Center for Interacting Urban Networks, through the NYUAD Research Institute Award [Grant CG001] and in part by the Swiss Re Institute through the QuantumCities Initiative.

The paper can be accessed here: https://pubsonline.informs.org/doi/abs/10.1287/trsc.2021.1058