On the interaction between Autonomous Mobility-on-Demand systems and the power network: models and coordination algorithms

Deep Reinforcement Learning-Based Charging Pricing for Autonomous Mobility-on-Demand System

“…Inspired by [29], we adopt a multi-commodity network flow model to capture the price responsive behavior of AMoD system. In this model, the AMoD operator is assumed to determine the order serving routes (i.e., order-serving), reposition idle vehicles for future order-serving and charging needs (i.e., vehicle repositioning), and schedule charging time and location for vehicles (i.e., charging scheduling) in a given transportation network, which is modeled in the following.…”

“…We consider a pricing scheme in a day-ahead manner so the AMoD operator can get those price signals in advance. It should be mentioned that the pricing scheme can be adjusted to real-time pricing, but in that case, the responsive model of AMoD operator also needs to be reformulated as receding horizon one [29]. In the proposed framework, each CSO makes pricing decisions in consecutive days d = 1, 2, • • • , D (called an episode).…”

“…The topology of the transportation system in the Dallas-Fort Worth metroplex is shown in Fig. 4, which is modified based on [29]. The targeting area is divided into 180 grids with 5 times 5 kilometer resolution, and the corresponding road network contains 30 nodes and 116 roads as simplified from OpenStreetMap [40].…”

“…The EVCS parameters are listed in Table I. The daily traveling requests for AMoD are assumed as 390,000 based on [29]. The commuters' value of time and distance is set equal to $4.40/h and $0.1/km, respectively [29].…”

Congestion-Aware Routing, Rebalancing, and Charging Scheduling for Electric Autonomous Mobility-on-Demand System

“…The routing problem is a mixedinteger linear program, but the problem has a limit in that it is solvable only when the origin-destination pairs are given. Rossi et al [20] developed a network flow and considered fleet impact on the power network. Estandia et al [21] extended this work and additionally considered the power distribution network.…”