Implementing a multi-user framework for vehicle routing problems: a chronicle

We report the design and implementation of a framework for solving general instances of vehicle routing problems, which was implemented within a commercial software for vehicle tracking and routing. The implementation of optimization algorithms in multi-user environments involves challenges that differ from the implementation of ad-hoc algorithms for particular problems, since a single implementation must accommodate the needs of several users and must be able to tackle instances with different characteristics. We review our first implementation and the extensions that were incorporated in successive versions, motivated by user needs. We comment on the experience and issues raised by a long-term relationship between a research team and a commercial company. Finally, we review the lessons learned along the way, in particular concerning algorithm design, coding practices, and research possibilities.