A Large Neighborhood Search-based approach to tackle the very large scale Team Orienteering Problem in industrial context

The Team Orienteering Problem (TOP) is an optimization problem belonging to the class of Vehicle Routing Problem with Profits in which the objective is to maximize the total profit collected by visiting customers while being limited to a time limit. This paper deals with the very large scale TOP in an industrial context. In this context, computing time is decisive and classical methods may fail to provide good solutions in a reasonable computational time. To do so, we propose a Large Neighborhood Search (LNS) combined with various mechanisms in order to reduce the computational time of the method. It is applied on classical sets of instances from the literature and on a new set of very large scale instances ranging from 1001 to 5395 customers that we adapted from Kobeaga et al. (2017). On the small scale set of instances, most best-known solutions are found. On the large scale set of instances, three new best-known solutions are found while the algorithm quickly gets more than half of the other best-known solutions.