Solving vehicle routing problems with OpenStreetMap and VROOM

Sunday 10:30, S.0.2

Julien Coupey 30 minutes


VROOM is an open-source software written in C++ to solve vehicle routing problems (VRP) arising in logistics and more widely in any context with geographically distributed tasks. Using OpenStreetMap data to solve real-life problems, VROOM offers an out-of-the-box integration with OSRM and is extensible to work on top of any routing engine.

The solving approach aims at providing high quality solutions efficiently by using dedicated (meta-)heuristics. This allows to get solutions very fast and/or to scale to huge problem sizes. Testing against TSPLIB, the reference benchmark for the Traveling salesman problem, shows an average optimal gap of only +2.47% while solving most instances in a few (milli)seconds.

We will describe the different ways to use VROOM by going through sample use-cases, and present the recent work toward supporting more VRP variants.