Abstract
The first-mile transportation system connects scattered requests in residential areas to mass transit networks and provides convenient and high-quality travelling services. The complicated road network and passenger mobility lead to potential multiple selections of passengers when choosing pick-up locations, which is ignored in the literature. This study develops a multi-objective optimisation model for the first-mile transportation system considering requests' preference ranks for multiple boarding stops. The objectives are to minimize the system cost and the number of unserved requests as a proxy for service quality, respectively. We devise a hybrid solution approach combining the fast elitist non-dominated sorting genetic algorithm (NSGA II) with a variable neighbourhood search (VNS) improvement method. The method is tested and compared with the classical NSGA II and an exact method on a number of instances. We also examine the effect of multiple stops and the discount scheme on the system performance.
Disclosure statement
No potential conflict of interest was reported by the author(s).