A memetic algorithm for a real-world dynamic pickup and delivery problem

The dynamic pickup and delivery problem (DPDP) is essential in supply chain management and logistics. In this study, we consider a real-world DPDP from daily delivery scenarios of a company. In the problem, orders are generated randomly and released periodically. The orders should be completed as soon as possible to minimize the cost. We propose a novel memetic algorithm (MA) to address this problem. The proposed MA consists of a genetic algorithm and a local search strategy that periodically solves a static pickup and delivery problem when new orders are released. We have conducted extensive experiments on 64 real-world instances to assess the performance of our method. Three state-of-the-art algorithms are chosen as the baseline algorithms. Experimental results demonstrate the effectiveness of the MA in solving the real-world DPDP.