This repository contains the data and resources used in our comprehensive review of methods that combine Metaheuristics (MH) and Reinforcement Learning (RL) to solve the Vehicle Routing Problem (VRP).
The VRP is one of the most well-known and challenging problems in logistics and combinatorial optimization. It involves determining optimal routes for a fleet of vehicles to serve a set of customers while satisfying various constraints, such as vehicle capacity and operational limitations.
Over the years, researchers have addressed this problem using a wide range of approaches, including exact methods, heuristics, metaheuristics, and machine learning techniques.
In this work, we focus on the emerging area of hybrid MH–RL approaches, which leverage the strengths of both paradigms:
- Metaheuristics for efficient exploration of large search spaces
- Reinforcement Learning for adaptive and data-driven decision-making
These hybrid methods are gaining increasing attention due to their ability to produce high-quality solutions efficiently.
This repository provides:
- A comprehensive review of advanced techniques for solving the VRP
- A focus on hybrid approaches combining Metaheuristics and Reinforcement Learning
- A structured classification of the literature into three main categories
- Insights into emerging trends, current challenges, and future research directions
-
All_Articles.xlsx
Contains the full dataset of reviewed articles -
Bibliographical_analysis-main.zip
Includes the code used to perform the bibliographical analysis -
Analysis.xlsx
Stores the analytical results obtained in the study -
MH+RL-Journals.png
Visualization of journal distribution for hybrid MH+RL studies -
MH-Journals.png
Visualization of journal distribution for metaheuristics-focused studies -
RL-Journals.png
Visualization of journal distribution for reinforcement learning studies -
Metaheuristics_Analysis_Results.txt
Provides a ranked list of metaheuristics based on frequency, along with their aggregated values -
README.md
This file, providing an overview and usage guide for the repository
The main objective of this review is to provide a clear and structured understanding of how Metaheuristics and Reinforcement Learning can be effectively combined to solve the VRP.
By analyzing existing work, we aim to:
- Highlight key challenges in the field
- Identify research gaps
- Provide insights into promising future directions
Hybrid MH–RL approaches represent a promising research direction in optimization and intelligent decision-making. By combining learning capabilities with powerful search strategies, these methods open new possibilities for solving complex and large-scale routing problems.
We hope this repository will serve as a valuable resource for researchers and practitioners working in:
- Combinatorial optimization
- Logistics and transportation
- Machine learning and reinforcement learning
For questions or collaborations, feel free to reach out.