Genetic algorithms for vehicle routing problem in delivery system

Abstract

Genetic algorithms are proposed as a new learning paradigm for combinatorial optimization that models a natural evolution mechanism. The authors attempt to apply genetic algorithms to the vehicle routing problem. As it is easy to generate the same gene while a generation shift goes on, it is feared that a solution will fall into a local minimum. The authors propose a new method that does not permit overlapping of genes. Some experiments are performed on digital road maps. The authors' results show that the genetic algorithms can effectively find optimum solutions.<>