Dynamic strategies optimizing benefits of fully autonomous shared vehicle fleets

Abstract

A constrained optimization framework of a flexible demand responsive transport system is considered. An intelligently administered scheme consisting of unmanned vehicles, requiring no prior seat reservation is introduced ensuring high quality door-to-door services at reduced costs. A decentralized decision making scheme comprised of various model based adaptive control patterns is developed. At any time optimized use of the available vehicle capacity is achieved while keeping cars as busy as possible. Vehicle itineraries are smartly defined according to their current state, traffic conditions and demand as well customer preferences. Tolerated passenger detours are respected while taking into consideration the related client waiting time. The asynchronous system behavior is modeled based on theory and methodology of discrete event dynamic systems (DEDS). Discrete event simulations permit evaluation of the system performance as well optimal tuning of the involved control algorithms. After identification of the desirable DEDS states the system is guided to controllable events infinitely often. As a case study, the city of Paris is considered. A comparative study is conducted appraising the suggested vehicle fleet versus a scheme consisting of self-service autonomous vehicles (SSAV). Metrics on cars, clients and network are presented such as trip durations, client waiting time and queue lengths at nodes, vehicle occupancy etc.