Adapted Genetic Algorithm for Orienteering Problem

The Orienteering Problem (OP) is a well-known route optimization problem with various real-world applications. It aims at finding the most valuable path by visiting a set of nodes within a given path length limit. It involves both the node selection and the path optimization and thus is difficult to solve. Genetic algorithm (GA) is known for its effectiveness in solving combinatorial optimization problems. Inspired from this, this paper adapts the genetic algorithm to solve OP mainly by adapting the crossover operation. To this end, this paper adapts seven popular crossover operations, namely position-based crossover based on order selection (OSPBX), position-based crossover based on greedy selection (GSPBX), order crossover based on order selection (OSOX), order crossover based on greedy selection (GSOX), partially mapped crossover (PMX), heuristic crossover (HX), and sequential constructed crossover (SCX). To further improve the quality of the obtained solution, this paper first leverages the 2-opt method to fine-tune the path and then further devises a vertex insertion method to additionally add more nodes in the fine-tuned path when its travelling length is within the given limit. Experiments have been conducted on various OP instances from small-scale to large-scale to investigate the effectiveness of the adapted GA with the seven crossover operators. The experimental results reveal that the adapted GA with seven crossover operators are very promising for solving OP.