Solving the Team Orienteering Problem with Cutting Planes

TL;DR

This paper presents a novel cutting plane-based algorithm for the Team Orienteering Problem, effectively solving larger instances and proving optimality for previously unsolved cases.

Contribution

It introduces new methods like arc relevance, customer mandatory inclusion, and advanced cuts, enhancing the linear formulation approach for TOP.

Findings

Proved optimality for 12 previously unsolved instances.

Algorithm is competitive with existing methods.

Effective use of polynomial variables and cutting planes.

Abstract

The Team Orienteering Problem (TOP) is an attractive variant of the Vehicle Routing Problem (VRP). The aim is to select customers and at the same time organize the visits for a vehicle fleet so as to maximize the collected profits and subject to a travel time restriction on each vehicle. In this paper, we investigate the effective use of a linear formulation with polynomial number of variables to solve TOP. Cutting planes are the core components of our solving algorithm. It is first used to solve smaller and intermediate models of the original problem by considering fewer vehicles. Useful information are then retrieved to solve larger models, and eventually reaching the original problem. Relatively new and dedicated methods for TOP, such as identification of irrelevant arcs and mandatory customers, clique and independent-set cuts based on the incompatibilities, and profit/customer restriction on subsets of vehicles, are introduced. We evaluated our algorithm on the standard benchmark of TOP. The results show that the algorithm is competitive and is able to prove the optimality for 12 instances previously unsolved.