A comparison of multiple objective algorithms in the context of a dial a ride problem

Abstract

In their operations private chauffeur companies have to solve variations of the multiple objective dial a ride problem. The number and type of restrictions make the problem extremely intricate and, when manually done, requires specialized people with a deep knowledge of the modus operandi of the company and of the environment in which the procedure takes place. Nevertheless, the scheduling can be automated through mean of computational methods, allowing to deliver solutions faster and, possible, optimized. In this context, this paper compares six algorithms applied to solving a multiple objective dial a ride problem, using data from a company mainly working in the Algarve, Portugal. The achieved results show that $$\epsilon $$-MOEA overcomes the other algorithms tested, namely the NSGA-II, NSGA-III, $$\epsilon $$-NSGA-II, SPEA2, and PESA2.