Shared autonomous electrified vehicle (SAEV) operation and system design optimization

Abstract

Shared autonomous electrified vehicles (SAEVs) combined with autonomous driving technology, car-sharing services, and electrified vehicles are expected to transform the transportation systems in the near future. This study presents three research subjects related to operation and system design optimization of SAEVs. Since battery electric vehicles (BEVs) emit less environmental pollution and have higher energy efficiency than internal combustion engine vehicles (ICEVs), shared autonomous battery electric vehicles (SABEVs) using BEVs as electrified vehicles of SAEVs have been developed. However, existing studies on the operation and optimization of SABEV systems do not take into account uncertainties in the SABEV systems, resulting in performance or objective that is far from the intended value. Therefore, the first research considers the uncertainties of the SABEV system and introduces reliability-based design optimization (RBDO) to the design of the SABEV system to propose a design framework for SABEV system that can minimize the total cost of system design while securing the reliability on the customer wait time. Another crucial issue on operation and optimization of SABEV systems is the movement of idle vehicles

however, most existing SABEV studies use randomly moving strategies, resulting in inefficient fleet operation. Once the predicted passenger demands are given, the efficiency of fleet operation can be improved by moving the vehicle in advance to a location where passenger demands are expected. Therefore, the second research proposes a deep learning-based idle vehicle relocation strategy that uses deep learning to predict passenger demand and approximate the optimization process required in the relocation strategy to perform idle vehicle relocation in real-time. Along with BEVs, fuel cell electric vehicles (FCEVs) are considered as next-generation vehicles, and it is easy to envision shared autonomous fuel cell electric vehicles (SAFCEVs) using FCEVs as electrified vehicles of SAEVs. Therefore, the third research focuses on extending the RBDO framework for SABEV system presented in the first research and deep learning-based idle vehicle relocation strategy proposed in the second research to the design framework for SAFCEV system.