3-opt Metaheuristics for the Probabilistic Orienteering Problem

Abstract

The Probabilistic Orienteering Problem (POP) is an optimization problem arising in logistics. A set of customers, each with a probability of requiring a service and a price to be collected in case the service is provided, is given together with deterministic travel times between customers. Given a time budget (length of the delivery window), the problem is to select a subset of the customers to be served within the time budget, in such a way that maximize the expected total prize collected, while minimizing the total expected travel time. The use of a 3-opt heuristic routine to carry out the optimization is discussed in this paper. In particular, it is investigated how such an approach can benefit from the use of a Tabu Search paradigm, and how the best results achieved compared with the state-of-the-art. A vision on how the 3-opt heuristic can improve the speed and efficiency on certain classes of POP instances is given.