Solving a bus driver scheduling problem with randomized multistart heuristics

Abstract

The Bus Driver Scheduling Problem is an extremely complex part of the Transportation Planning System (e.g., [12]) that generally can be divided in different subproblems due to its complexity: Timetabling, Vehicle Scheduling, Crew Scheduling (Bus and Driver Scheduling), and Crew Rostering (see Figure 1 for the relationship between these subproblems). The transportation service is composed of a set of lines that corresponds to a bus traveling between two locations of the same city or between two cities. For each line, the frequency is determined by the demand. Then, a timetable is constructed, resulting in journeys characterized by a start and end point, and a start and end time. The Vehicle Scheduling Problem consists in finding a schedule for the buses, each schedule being defined as a bus journey starting at the depot and returning to the same depot. The objective is to minimize the total cost given by the cost of buses used for the service and running costs. Running costs can be minimized avoiding unnecessary deadheads, i.e., trips carrying no passengers. The daily schedule of each single bus is known as a running board (or vehicle block).