Advances in AI-Based Solutions for the Multi-Depot Vehicle Routing Problem: A Review of Recent Trends and Future Directions
The Multi-Depot Vehicle Routing Problem (MDVRP) is a complex combinatorial optimization challenge that involves determining the most efficient routes for a fleet of vehicles operating from multiple depots while considering constraints such as capacity, time windows, and customer demands. Due to its NP-hard nature, MDVRP presents significant difficulties in finding optimal solutions. In recent years, artificial intelligence (AI) techniques, particularly metaheuristic algorithms inspired by natural processes and deep learning (DL) and reinforcement learning (RL) methods, have gained prominence for their ability to offer scalable and efficient solutions to MDVRP. Integrating hybrid approaches combining metaheuristics with DL has been especially noteworthy, providing enhanced solution quality and computational efficiency. Studies have shown that hybrid optimization strategies, such as clustering techniques with genetic algorithms (GAs), optimize customer assignments and overall route planning. This paper systematically reviews recent literature from the past ten years to identify common AI-based approaches for addressing MDVRP. It discusses potential directions for future research to advance the field further. Future research should focus on real-time data integration through IoT, which can create more dynamic and adaptive solutions. Additionally, AI methodologies, such as RL algorithms paired with metaheuristics, hold the potential for addressing larger, real-world MDVRP instances. Applying AI to practical domains like e-commerce logistics, health logistics, and emergency response has demonstrated significant operational improvements. Continued exploration of multi-objective optimization that balances cost, route balance, and energy efficiency is crucial for handling real-world complexities, including demand and capacity uncertainties.
- Journal
- ARCHIVES OF COMPUTATIONAL METHODS IN ENGINEERING
- Year
- 2026
- Volume
- 33
- Issue
- 2
- Pages
- 1997-2035
- ISSN
- 1134-3060
- eISSN
- 1886-1784
- DOI
- 10.1007/s11831-025-10356-y
- WoS UID
- WOS:001577142400001