The Performance of Genetic Algorithms on Solving the Pickup and Delivery Vehicle Routing Problem with Soft Time Windows

Abstract

This study examines the performance of genetic algorithms (GAs) on solving the pickup and delivery vehicle routing problem with soft time windows (PDVRPSTW), typically faced by home delivery service providers. We implement a number of combinations of GA's three main algorithmic operators, namely, selection, crossover, and mutation, and adopt the data envelopment analysis (DEA) to evaluate and rank these various combinations of GA operators. The numerical results show that DEA is appropriate in determining the efficient combinations of GA operators. Among the combinations under consideration, the combinations using tournament selection and simple crossover are generally more efficient. The findings also contribute to algorithm development and evaluation in vehicle routing problem from the operational research perspective.