Collective cars in cities: Optimal management in real time through DES approach

Abstract

A decentralised management of an autonomous demand responsive transport mode is considered, where more than one passengers may be associated to a vehicle. No advanced seat reservation is required by clients appearing in the network in a continuous Poisson stream aiming at an immediate departure. A potential waiting time is associated with customers, in case no vehicle is available by their arrival, as well a tolerated trip detour threshold. Meanwhile, vehicles travel on the network transporting passengers while searching for prospective clients. Two different schemes, employing fully autonomous and manually driven cars are considered. A comparative study is presented, quantifying the performance of each structure. Moreover, the efficacy of each vehicle scheme is compared with a similar one employing a greater number of vehicles in service where each vehicle has an increased passenger capacity. Client-vehicle interaction forms an asynchronous, dynamical, spatio-temporal system the evolution of which is represented through an event-driven approach. Client metrics such as client initial waiting time, abandonment rate, trip detours etc. are measured as well vehicle indicators providing information on the vehicle productivity.