Reliability-constrained multimodal journey planning for students: a bi-objective empirical evaluation

Journey planning based on published timetables often fails for time-critical trips because public transport operations deviate from schedule. This article proposes reliability-constrained multimodal routing, where empirical event-level arrival and departure bounds are enforced as transfer feasibility constraints during route construction. Bounds are estimated from six months of observations and embedded in a time-expanded network combining public transport, walking, personal cycling, and bikeshare. A bi-objective backward search maximises the latest feasible departure time while minimising monetary cost, with planning parameters calibrated from a survey of 958 students prioritising reliability and travel time. The approach is evaluated on 403 commuting queries in Strasbourg over five held-out days under public transport, public transport with bikeshare, and multimodal configurations. Under the most conservative setting, late arrivals decrease from 64.2 to 2.6 with public transport, from 88.4 to 3.2 with bikeshare, and from 7.0 to 0.2 in the multimodal configuration.