Modeling autonomous dynamic vanpooling services in SUMO by integrating a dynamic routing scheduler

Dynamic vanpooling services, as a flexible, demand adaptive transport service with ridesharing, is rather a complex transport mode. To model such a service requires a similar behavior/environment of dynamic traffic simulation environment linked with a dynamic rerouting algorithm to control the service vehicles serving stochastic requests and incurring the time-dependent network state. We develop a modeling platform for autonomous dynamic vanpooling services based on the traffic simulator SUMO, using its TraCI’s API by integrating a dynamic routing scheduler, along with the required simulation setup to model such ridesharing vans. The integrated routing algorithm solves the rerouting optimization problem for both stochastic requests and time-dependent travel times, hence enabling the developed platform to model dynamic vanpooling services stochastically (i.e. stochastic demand requests arrival and service optimization during the simulation execution). A simulation executer/controller module, written in python, loading TraCI API library and the routing algorithm (C++ code wrapped as a dll) is developed controlling the simulation execution integrated with the routing scheduler by predefined communication protocols through TraCI. The platform can provide benefits to model and evaluate dynamic vanpools for understanding their feasibility, behavior, and impacts for different situations (i.e. networks/service area, network conditions, demographics, peak-offpeak hours, etc.). The platform implementation is still in the final phases and some preliminary results are depicted for different van fleet sizes and passenger requests served, running the scheduler based on dynamic travel time information linked from SUMO.