Real-time passenger bus routing problems with preferences and tradeoffs

One category of vehicle routing problems involving groups of people where there can be multiple possible drop-off locations is the School Bus Routing Problem (SBRP). In this paper, we investigate a form of SBRP where each passenger has a well-ordered set of preferred stops, e.g., the preferences can be based on the distance to their destination. The passengers continue their journey from the stop to their destination by an alternative mode, such as walking. Different drop-off choices allow for a different set of passengers to be dropped off together, affecting the total vehicle driving distance. As the transport provider and passengers have their own objectives, inevitably, there are two opposing goals (tradeoff aspects), reducing total driving distance versus passenger walking distance. Unlike typical SBRP, we focus on practical problems where the set of passengers is not known well in advance and there is only a short time window to compute a solution (a real-time requirement). Hence, runtime efficiency is more important than optimality. We introduce this variant as SBRP-RT, a school bus routing problem with tradeoff aspects and real-time requirements. We propose a model with an efficient greedy local search for such problems, and demonstrate the efficacy with experiments on real-world datasets with quick solving that balances tradeoffs.