Multi-objective memetic algorithm for solving pickup and delivery problem with dynamic customer requests and traffic information

Jun Xiao, Yanming Yang, Xiaoliang Ma, Jiarui Zhou, Zexuan Zhu

Abstract: This paper formulates one-to-many-to-one pickup and delivery problems with dynamic customer requests and traffic information. A multi-objective memetic algorithm namely prioLSH-MOMA is proposed to solve the problems. The new algorithm is characterized with a priority and locality-sensitive hashing based local search. prioLSH-MOMA is designed to find an optimal route of a dynamic pickup and delivery problem in terms of route length and workload. Particularly, a re-planning strategy is introduced to handle the dynamic information. Priority and locality-sensitive hashing based local search is applied to fine-tune the candidate routes during the evolution process. prioLSH-MOMA is evaluated with two dynamic pickup and delivery problems simulated on real-world maps and the results demonstrate the efficiency of the proposed algorithm.