A Four-stage Heuristic Algorithm for Solving On-demand Meal Delivery Routing Problem

TL;DR

This paper introduces a four-stage heuristic algorithm for on-demand meal delivery routing that efficiently combines orders, optimizes routes, and assigns couriers, improving service quality and reducing travel distance in real-time scenarios.

Contribution

The paper presents a novel four-stage heuristic algorithm that integrates order bundling, route optimization, and courier assignment for meal delivery routing problems.

Findings

The algorithm reduces average delivery time and travel distance.

It outperforms existing algorithms in solution quality.

The method effectively handles real-time data for dynamic routing.

Abstract

Meal delivery services provided by platforms with integrated delivery systems are becoming increasingly popular. This paper adopts a rolling horizon approach to solve the meal delivery routing problem (MDRP). To improve delivery efficiency in scenarios with high delivery demand, multiple orders are allowed to be combined into one bundle with orders from different restaurants. Following this strategy, an optimization-based four-stage heuristic algorithm is developed to generate an optimal routing plan at each decision point. The algorithm first generates bundles according to orders' spatial and temporal distribution. Secondly, we find feasible bundle pairs. Then, routes for delivering any single bundle or multiple bundles are optimized, respectively. Finally, the routes are assigned to available couriers. In computational experiments using instances from open datasets, the system's performance is evaluated in respect of average click-to-door time and ready-to-pickup time. We demonstrate that this algorithm can effectively process real-time information and assign optimal routes to the couriers. By comparing the proposed method with existing the-state-of-the-art algorithms, the results indicate that our method can generate solutions with higher service quality and shorter distance.