A cooperative strategy for optimizing vehicle relocations and staff movements in cities where several carsharing companies operate simultaneously

Communication‐aware distributed rebalancing for cooperative car‐sharing service

“…However, if numerous providers can share their car stations, it is anticipated to reduce the costs by enhancing the effectiveness of vehicle operation. The rebalancing control strategies for cooperative car-sharing service have also attracted great attention, including an integer programming formulation [10][11][12][13][14], a method on Markov chain modeling [15], a sparse modeling method [16], a game-theoretic method [17], and a learning-based approach [18]. A deep-learning method of cooperative car-sharing service was also proposed to optimize shared vehicles and staff drivers over several different providers [19].…”

“…The rebalancing control strategies for cooperative car-sharing service have also attracted great attention, including an integer programming formulation [10][11][12][13][14], a method on Markov chain modeling [15], a sparse modeling method [16], a game-theoretic method [17], and a learning-based approach [18]. A deep-learning method of cooperative car-sharing service was also proposed to optimize shared vehicles and staff drivers over several different providers [19]. The information on all service providers and stations is collected to a data center and processed and centralized in most of these methods.…”

“…• The distributed algorithm for the optimal rebalancing control problem is proposed so that each provider can obtain the optimal operation of the deadhead vehicles in a distributed manner without disclosing the detailed data on the own constraints and the demand data to other providers. Therefore, compared with the centralized methods in [10][11][12][13][14][15][16][17][18][19], the obtained solution is suitable for large-scale sharing services that require more scalability and anonymity. • The local communication between service providers is conducted only when necessary.…”

“…where (18b) and (18c) are the dynamics of the car-sharing service and the initial condition corresponding to (13), and (18d)-(18f) are the constraints corresponding to ( 14)- (16). The cost function of provider i that represents the cost of the deadhead vehicles is presented as…”

“…To solve the finite-time optimal control problem (18) in a distributed manner, we reformulate the problem (18) as an optimization problem with a coupling constraint. From (13), the dynamics of the car-sharing service of provider i is given as…”