A multi-objective memetic algorithm based on locality-sensitive hashing for one-to-many-to-one dynamic pickup-and-delivery problem

This paper presents an early attempt to solve one-to-many-to-one dynamic pickup-and-delivery problem (DPDP) by proposing a multi-objective memetic algorithm called LSH-MOMA, which is a synergy of multi-objective evolutionary algorithm and locality-sensitive hashing (LSH) based local search. Three objectives namely route length, response time, and workload are optimized simultaneously in an evolutionary framework. In each generation of LSH-MOMA, LSH-based rectification and local search are imposed to repair and improve the individual solutions. LSH-MOMA is evaluated on four benchmark DPDPs and the experimental results show that LSH-MOMA is efficient in obtaining optimal tradeoff solutions of the three objectives.