Multi-period street scheduling and sweeping

Abstract

In the street-sweeper problem, we seek to sweep the sides of city streets in a way that minimises the distance travelled by the street sweepers. Typically, street sweepers are blocked by parked cars that prevent the curb from being swept. We consider a problem posed to us by Washington, DC where the parking constraints are multi-period decision variables. For example, suppose the city wishes to sweep its streets over two days with available parking on at least one side of each street on each day. Before the city considers how to sweep its streets, it must first decide which street sides to make available for parking on each day in a way that obeys the parking constraint on both days. We present a genetic algorithm that generates high-quality solutions and discuss managerial implications.