MA-ABC

Abstract

Services such as garbage collection, road gritting, street sweeping, and power line inspection can each be formulated as a capacitated arc routing problem (CARP). The traditional formulation of CARP has the goal of minimizing the total cost of the routes making up a solution. Recently, operators of such services require routes that are balanced and visually attractive in addition to low cost. Routes that are balanced are about equal in length and provide fair work assignments. Visually attractive routes are subjective, but they usually involve non-crossing routes that provide well defined service areas. These additional features are important because they address operational complexities that arise from using the routes in practice. This paper presents MA-ABC, a memetic algorithm to find solutions for CARP that maximize route attractiveness and balance, while minimizing total cost. A novel fitness function combines route overlap with route contiguity to assess route attractiveness. MA-ABC is the first to incorporate attractiveness in a three-objective search for heuristic solutions for CARP. Experimental results on CARP benchmark instances show that MA-ABC finds a diverse set of heuristic solutions at the Pareto front, providing a wide choice for service operators to tradeoff design objectives.