A real geographical application for the School Bus Routing Problem

This work presents our research and implementation of the School Bus Routing Problem applied to the rural area of a Brazilian city. We use a complete set of real georeferenced data containing a sample of 944 students, 23 schools, and the full road network of a city of population of 280,000 inhabitants occupying an area of 2,348 km
²
. Our goal is to optimize the daily transportation of students considering the real publicly available fleet composed of heterogeneous vehicles. As our solution is being evaluated for application over the entire country, we have included some 'quality' parameters for tunning the application to distinct economic/social realities. A mixed integer linear programming model and a GRASP based algorithm are proposed to solve the routing problem.