Minimizing last-mile emissions through altitude-aware route optimization

Abstract

This study introduced an exact optimization approach to solve a new special type of Travelling Salesperson Problem. This problem considers time-windows restrictions and a new objective function—the objective regards driver assessment awareness and fuel consumption. The latter is modeled as a function of variable vehicle payload and terrain elevation data. This problem can be stated as the way to find the best route to service a set of customer demands, attempting to deliver within agreed time windows, mimicking paths that are as similar as possible to good routes executed in the past by experienced drivers, allowing small alterations as to reduce duration and fuel consumption. The authors proposed an innovative mixed integer linear programming formulation and a cluster decomposition approach that reduces search space and makes the approach applicable to solving real-world-sized problems. This model was parametrized using a small-sized mockup dataset and had its applicability tested on real data. The latter consisted of a public dataset containing trips executed and evaluated by real drivers of Amazon company. The results show that it was possible to reduce in -5.7% the fuel consumption in the routes of this dataset. Since this variable is directly related to emissions and pollution, this result shows that the suggested approach offers promising prospects for improving efficiency and reducing the carbon footprint of logistics last-mile operations. To the best of the authors' knowledge, this study contributes to the literature in that it is the first to jointly tackle driver assessment awareness and fuel consumption in a route optimization problem. Thus, it is also the first to propose a mathematical formulation and solution approach for this problem.