Dynamic Vehicle Allocation and Charging Policies for Shared Autonomous Electric Vehicles

René De Koster,Yuxuan Dong,Debjit Roy

2019-01-01

Abstract:Vehicle sharing systems, which are considered to be a sustainable solution for urban transportation, can nowadays be found in many cities (He et al., 2017). In New York City, for example, there are three types of vehicle sharing services. The first, also the most popular, is ride-sharing, where a platform coordinates the rides shared by drivers with customers who need a ride. As shown in Figure 1(a), the drivers pick up the customers at their origins and drop off them at their destinations. Since the drivers instead of the platform (e.g., Uber) own the vehicles, the drivers may reject the match proposed by the platform, which may increase the overall driving distances (Wang et al., 2017). Free-floating vehicle sharing is another vehicle sharing service (e.g., car2go). Customers choosing this service have to first search a vehicle in their proximity, go to it, and then drive to their destinations. This is shown in Figure 1(b). The vehicle can be left at any parking space at the destination. However, the company offering this service needs to hire a crew to reposition and to refuel the vehicles for maintaining daily operations (Perlman, 2014). Station-based vehicle sharing is a service where vehicles must be picked up and left at designated refueling or recharging stations (shown in Figure 1(c)). Since the platform owns all vehicles, customers must drive vehicles themselves in both freefloating and station-based vehicle sharing systems. To attract more customers to use these two types of services, more vehicles need to be deployed and more stations to be built in the service region, which lowers the utilization of the vehicles. Email addresses: rkoster@rsm.nl (René De Koster), dongyx@mail.ustc.edu.cn (Yuxuan Dong), debjit@iima.ac.in (Debjit Roy) Preprint submitted to TSL2020 December 18, 2019 Figure 1: Four types of vehicle sharing services Autonomous electric vehicle (AEV) sharing systems can reduce the disadvantages of the three vehicle sharing services (Daimler, 2017; Jing, 2017). From the customer perspective, it is similar to ride-sharing. The only difference between ride-sharing and AEV sharing is that the latter does not need drivers (illustrated in Figure 1(d)). From the platform perspective, the platform (or another company) owns the vehicles, which are autonomous and electric. In an AEV sharing system all processes, especially matching vehicles with customers and charging vehicles, need to be controlled by the platform. Moreover, customer patience times may vary with travel distances. In practice, longdistance customers can endure a longer waiting time than customer who wish to travel a short distance. If the waiting times of customers exceed their maximal patience time, they will leave the system and choose for other transport options. Loosing a long-distance customer implies revenue loss. Therefore, operating an AEV sharing system introduces a new challenge: how should the platform allocate vehicles to customers and when to charge them? There are some differences with manned vehicle sharing systems. In ride-sharing, there is no need to consider refueling or recharging because the platform does not own the vehicles (Bai et al., 2018). For free-floating and station-based vehicle sharing services, refueling or recharging is executed manually, and also outside the control of the platform. Therefore, in literature on these systems, activities related to charging are usually simplified as exogenous parameters rather than endogenous decisions (He et al., 2017; Bai et al., 2018). In the studies on vehicle routing and recharging, charging times and paths