On-Demand Dynamic Intercity Shared-Taxi System With Re-Optimization

This paper develops a forward-looking system for private companies, providing real-time intercity rideshare services on demand for passengers traveling between two cities, using a fixed homogeneous fleet to respond instantly to every request. A major challenge faced by the system is that adopting the algorithmic framework of urban ridesharing could lead to extremes under the context of longer intercity travel distances and higher vehicle capacities: 1. Dynamic reassignment leads to frequent reassignment of passengers, constantly changing real-time route information, deteriorating the travel experience for both drivers and passengers; 2. Lack of dynamic reassignment results in myopic, reducing corporate profits. In response, this paper proposes the Sharebility Graph-based Restricted Adaptive Large Neighborhood Search (SG-RALNS) algorithm for dynamic reassignment of vehicle routes. The core idea is to balance passenger experience and corporate profits by restricting reassignment parameters. Additionally, this work integrates the application of the Sharebility Graph (SG), including the design of insertion and deletion operators tailored for intercity travel. The performance of the algorithm is validated through real-world datasets, with quantitative results showing our algorithm can increase company profits by around 1% to 48%. Experimental results indicate that our algorithm framework is also applicable to other long-distance, high-capacity ridesharing scenarios, such as rural ride-sharing.