RUG: A Revenue-driven User-based Relocation Approach in One-way Car-sharing Incorporating Public Transportation

Car-sharing services can meet the diverse travel needs of users while helping to alleviate traffic congestion and reduce pollution. In many scenarios, car-sharing is more economical than taxis. One-way car-sharing allows users to rent and return cars at any station within the system, which leads to low operating costs and flexible services. However, the spatiotemporal skewness of user travel demand gives rise to imbalances between vehicle supply and demand among stations, which limits the profitability of car-sharing companies. Relocating vehicles can alleviate the above problems to some extent. Most existing studies construct optimization models with the goal of maximizing expected revenue or reducing system imbalance. The former is limited by the insufficient accuracy of travel demand prediction, and the mode of discarding definite orders and pursuing higher possible expected benefits instead cannot guarantee actual profits. To improve system balance, the latter pays more relocation costs such that reduces the profitability. To this end, we propose a revenue-driven one-way car-sharing user relocation model RUG that is suitable for real-time scenarios. The model is based on the deterministic effect of prospect theory, which ensures the current definite gains. For orders that cannot be fulfilled due to imbalanced resources, RUG provides users with alternative travel routes, which not only attempts for promising gains but also effectively balances the system. Users are incorporated into the system as relocation subjects by designing rational user incentive and acceptance models. Public transportation is utilized to break through the distance limitations of user relocation. Relocation plans are evaluated with a greedy heuristic. Experimental results on real-world New York datasets show that the RUG model has significant advantages over existing user-based relocation methods. Under the same parameters, compared to the representative user-based relocation method, RUG increases service order volume and profit by 14% and 60%, respectively. Notably, RUG can effectively raise unit profits during traffic rush hours. By incorporating travel demand forecasting, the model further increases revenue by 5.4% while also improving user service level and system balance. © 2023 Chinese Society of Agricultural Engineering. All rights reserved.