Vehicle routing problems are currently an important part of the decision-making process in logistic systems. Efficient routes bring benefits regarding transportation costs, quality of service, gas emission, work regulations, among other features. The literature on this subject has been active for decades and is characterized by a large number of publications every year, motivated by theoretical and practical challenges. However, the vast majority of contributions assume that the input data is accurate and known in advance, even though in practice we often find a completely different scenario. Routes are commonly affected by uncertainties caused by a variety of sources, such as car crashes, traffic jams, weather conditions and road/vehicle maintenance, which are completely overlooked by most publications. Ignoring these uncertainties at the route planning process is likely to result in ineffective or even infeasible routes in practice. To reduce this research gap, the objective of this project is to study and propose formulations and solution methods that incorporate data uncertainty into vehicle routing problems. We intend to address different paradigms, such as Stochastic Programming and Robust Optimization, and develop exact and heuristic methods that most suits each paradigm, considering scenarios and uncertainty sets obtained from probability distributions as well as historical data. Nowadays, companies are full of data that can be appropriately explored to improve the route planning process. This way, in addition to the proposition of new formulations and methods, we expect to contribute to the literature with a review and classification of the main approaches currently available to incorporate uncertainty into vehicle routing problems. Moreover, the proposed approaches have the potential of contributing to the improvement of decision-making processes in practice.
News published in Agência FAPESP Newsletter about the scholarship: