Section: New Software and Platforms

VRPsolve

Keywords: C++ - Mobile Computing, Transportation - Optimization

• Participants: Clive Ferret-Canape, Arnaud Liefooghe and Sébastien Vérel

• Contact: Arnaud Liefooghe

• URL: http://gforge.inria.fr/projects/vrpsolve (limited access)

Scientific Description

VRPsolve is a software for solving vehicle routing problems dealing with last-mile delivery issues that arise as we approach the final customer. When modeling and solving combinatorial optimization problems, especially problems related to the transport of goods and people, the resulting models are generally subject to a specific development in order to be validated, as industrial needs are highly dependent of the application domain. However, a set of conventional objectives and constraints, such as vehicles capacities, incompatible parcels, time windows, are now commonly encountered. In addition to being efficient and effective, VRPsolve differentiates from other tools by allowing to quickly and conveniently integrate ad-hoc constraints and objectives into a generic software. Indeed, VRPsolve effectively deal with industrial last-mile delivery vehicle routing problems and is able to cope with multiple objectives and a large number of constraints by using advanced optimization algorithms which are usually not available with existing softwares. In addition, VRPsolve allows industrial collaborations to be addressed by solving real-world problems requiring geographic information systems (GIS).

Functional Description

The current release includes the following functional and technical specifications:

• A modular architecture which allows for an easy integration into a global information system (with respect to data standards, weak coupling with external libraries, etc),

• An independent geographic information system with graphic display,

• A resolution engine based on metaheuristics (running in a high-performance computing mode),

• The possibility of coupling with other optimization solvers and frameworks like Paradiseo (in order to enable a quick prototyping of new optimization algorithms),

• Objective- and constraint-handling that can be easily configured by the user,

• Software engineering (build system testing, continuous integration, etc).