Genetic algorithms for the sailor assignment problem

This paper examines a real-world application of genetic algorithms – solving the United States Navy’s Sailor Assignment Problem (SAP). The SAP is a complex assignment problem in which each of n sailors must be assigned one job drawn from a set of m jobs. The goal is to find a set of these assignments such that the overall desirability of the match is maximized while the cost of the match is minimized. We compare genetic algorithms to an existing algorithm, the Gale-Shapley algorithm, for generating these assignments and present empirical results showing that the GA is able to produce good solutions with significant savings in cost. Finally, we examine the possibility of using the GA to generate multiple different solutions for presentation to a human decision maker called a detailer, and we show that the GA can be used to provide a sample of good solutions. Categories and Subject Descriptors

• Type: Conference
• Year: 2005
• Where: GECCO
• Authors: Deon Garrett, Joseph Vannucci, Rodrigo Silva, Dipankar Dasgupta, James Simien
• Keywords: Assignment Problem, GECCO 2005, Genetic Algorithms, Navy’s Sailor Assignment