An Integer L-shaped Method for Dynamic Order Fulfillment in Autonomous Last-Mile Delivery with Demand Uncertainty

Linxuan Shi,Zhengtian Xu,Miguel Lejeune,Qi Luo

2024-06-10

Abstract:Given their potential to significantly lower costs and enhance flexibility in last-mile delivery, autonomous delivery solutions like sidewalk robots and drones have garnered increased interest. This paper addresses the dynamic order fulfillment problem faced by a retailer who operates a fleet of low-capacity autonomous delivery vehicles, servicing requests arriving in a stochastic manner. These delivery requests may vary in package profiles, delivery locations, and urgency. We adopt a rolling-horizon framework for order fulfillment and devise a two-stage stochastic program aimed at strategically managing existing orders while considering incoming requests that are subject to various uncertainties. A significant challenge in deploying the envisioned two-stage model lies in its incorporation of vehicle routing constraints, on which exact or brute-force methods are computationally inefficient and unsuitable for real-time operational decisions. To address this, we propose an accelerated L-shaped algorithm, which (i) reduces the branching tree size; (ii) substitutes exact second-stage solutions with heuristic estimations; and (iii) adapts an alternating strategy for adding optimality cuts. This heuristic algorithm demonstrates remarkable performance superiority over the exact method, boasting a more than 20-fold improvement in average running time while maintaining an average optimality gap of less than 1%. It is then employed to solve a wide range of instances to evaluate the advantages of adopting the stochastic model. Our findings demonstrate long-term cost savings of up to 20% when accounting for demand uncertainty in order fulfillment decisions. Meanwhile, the derived savings could diminish as the uncertainty in order arrivals increases.

Optimization and Control

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to dynamically complete order allocation in the last - mile delivery in the presence of demand uncertainty. Specifically, the paper focuses on how retailers manage a fleet of low - capacity autonomous delivery vehicles (such as sidewalk robots or drones) to handle randomly arriving delivery requests. These requests may vary in package characteristics, delivery locations, and urgency levels. The paper uses a rolling time - window framework for order allocation and designs a two - stage stochastic programming model, aiming to strategically manage existing orders while considering future requests that may be affected by various uncertainties. The main challenge lies in how to incorporate vehicle routing constraints into the model, because exact or brute - force solution methods are computationally inefficient and not suitable for real - time operational decisions. To solve this problem, the paper proposes an accelerated L - shaped algorithm to improve computational efficiency by reducing the size of the branch tree, replacing exact second - stage solutions with heuristic estimates, and using an alternating strategy to add optimality cuts.