In this paper, a robust optimization model based on the adjustable decision rule is used to solve the empty vehicle scheduling problem for the operators of the vehicle-sharing system to explore potential users and achieve maximum profit, considering the impact of the uncertainty of passenger demand information on the scheduling process. Since the travel demands of passengers are uncertain in the vehicle-sharing system, this paper describes the corresponding uncertain set which limits the travel demands of passengers in a certain interval. Moreover, the sum of the passenger demands in time is limited flexibly to reduce the conservatism of the model. Subsequently, an adjustable decision rule is introduced to develop the empty vehicle scheduling strategy based on the observed demands, and a robust optimization model for empty vehicle scheduling is proposed as well as the tractable linear programming form of the model. By using real DiDi order information to simulate the travel demands of customers using shared cars, the numerical experiment shows that the empty vehicle scheduling strategy obtained by using the proposed model (compared with the deterministic model) will cost more on the empty vehicle scheduling, and make the operators achieve greater profits and meet the more demands of passengers both in average and in the worst case, which proves the robustness and practicability of the proposed model.