Robust timetable optimization for bus lines subject to resource and regulatory constraints

Timetables are typically generated based on passenger demand and travel time expectations. This work incorporates the travel time and passenger demand uncertainty to generate robust timetables that minimize the possible loss at worst-case scenarios. We solve the resulting minimax problem with a genetic algorithm that uses sequential quadratic programming to evaluate the worst-case performance of each population member. Our approach is tested on a bus line in Singapore demonstrating an improvement potential of ≃5% on service regularity and excessive trip travel times.