Design and development of an optimisation model and simheuristic framework for the on-demand delivery problem with driver-experience-based stochastic travel time

In this research, we investigate an on-demand delivery problem in city logistics, in which delivery requests from online customers are received in real-time and delivery services are conducted in short time. A multi-stage stochastic vehicle routing optimisation model is proposed, which incorporates two unique features arising from city logistics delivery, i.e., driver experience and stochastic travel time. In practice, the estimated travel time is indeed affected by the driver experience and can be gradually improved as drivers learn from accumulative delivery experience. A simheuristic framework is developed to handle the proposed model, in which an improved adaptive large neighbourhood search is designed for searching promising solutions with deterministic travel time, and Monte Carlo simulation is leveraged to assess the solution qualities and facilitate the searching process in stochastic scenarios. Computational experiments demonstrate that the proposed simheuristic framework can tackle the on-demand delivery problem with satisfactory performance. [Submitted: 22 November 2023; Accepted: 13 April 2024]