Vehicle routing problem (VRP) research has recently improved dramatically to simulate more real-life circumstances. Nevertheless, the typical VRP models proposed have been isolated from the most important factor determining the success of the VRP plan on the ground, i.e. the human factor (driver). Thus, this research investigates the effect of drivers' behaviours on the optimal VRP plan by integrating the level of autonomy of both planner and drivers as represented by risk-taking parameters. To enhance the model configuration's practicability, the concept of ‘ridesharing’ – which has been introduced before – has also been integrated to expand the logistical services, improve customer satisfaction, and compensate for shortages in service. Moreover, to ensure environmentally-friendly logistical practices, a velocity maximisation policy and environmental penalty enforcement on the chosen velocity range have been considered. In general, the model improves drivers' satisfaction, customers' perceived quality, and the firm's financial objectives. Additionally, it achieves a better supply chain strategic fit by planning at the three levels: strategic, tactical, and operational. A numerical example was solved using the Eclipse Java 2018-09 solver through two heuristic methods, the Greedy and the Intra-route neighborhood heuristic, and both revealed the same near-optimal solutions. Sensitivity analyses showed that the resulting insignificant increase in the VRP costs due to assigning autonomy for drivers is still reasonable, and the total costs' objective function weight has an insignificant effect on the total near-optimal solution, while that of the energy consumption objective function has the largest impact.
Read full abstract