Fast Multimodal Journey Planning for Three Criteria

Abstract: We study the journey planning problem for multimodal networks consisting of public transit and a non-schedule-based transfer mode (e.g., walking, bicycle, e-scooter). So far, all efficient algorithms for this problem either restrict usage of the transfer mode or Pareto-optimize only two criteria: arrival time and the number of used public transit trips. However, we show that both limitations must be lifted in order to obtain high-quality solutions. In particular, the time spent using the (unrestricted) transfer mode must be optimized as a third criterion. We present McTB, the first algorithm that optimizes three criteria efficiently by avoiding costly data structures for maintaining Pareto sets. To enable unlimited transfers, we combine it with a three-criteria extension of the ULTRA preprocessing technique. Furthermore, since full Pareto sets become impractically large for more than two criteria, we adapt an approach by Delling et al. to restrict the Pareto set in a methodical manner. Extensive experiments on real-world data show that our algorithms are fast enough for interactive queries even on large country-sized networks. Compared to the state of the art for multicriteria multimodal journey planning, MCR, we achieve a speedup of up to 80.