Genetic algorithm-based routing problem considering the travel reliability under asymmetrical travel time distributions

Travel time reliability is a critical factor affecting travelers' route choices, and standard deviation has been widely used as the quantitative measure in many routing models. However, there have been concerns regarding its behavioral interpretation and theoretical limitations when the underlying travel time distribution is asymmetrical. This study proposed use of semistandard deviation (SSD) as the measure of risk under uncertain conditions and demonstrates its conceptual advantages over its counterpart. Then, a routing strategy was formulated that minimized the total cost and that included both average travel time and a travel time reliability term between user-specified origin and destination. A sampling-based approach that used field-collected data was applied to capture the spatial dependencies of link travel times during the modeling process. The genetic algorithm was then adopted to solve the proposed model. Finally, the SSD-based model was numerically evaluated on a real-world network. The results indicate that the proposed model has a better interpretation of a traveler's route decision involving skewed travel time distribution with excessively long delays.