Scholarlink

<p>The Vehicle Routing Problem (VRP) is a critical challenge in supply chain logistics, aiming to optimize delivery routes while minimizing costs, travel time, and environmental impact. Traditional approaches, including exact algorithms, heuristics, and metaheuristics, struggle with scalability and adaptability in real-world logistics scenarios. The advent of artificial intelligence (AI) has revolutionized VRP optimization, particularly through reinforcement learning (RL) and graph neural networks (GNNs). RL formulates VRP as a sequential decision-making task, allowing adaptive and real-time route optimization. When combined with GNNs, which effectively model transportation networks as structured graphs, AI-driven methods enhance spatial awareness, optimize route selection, and improve computational efficiency. This study explores the integration of RL and GNNs in solving VRP and its key variants, including Capacitated VRP (CVRP), VRP with Time Windows (VRPTW), and Capacitated VRP with Time Windows (CVRPTW). The research presents a comparative evaluation of traditional versus AI-driven VRP solutions, analyzing performance metrics such as total travel distance, cost minimization, and computational efficiency. Experimental results demonstrate that the proposed AI framework outperforms conventional techniques by dynamically adapting to constraints and real-time disruptions. By leveraging AI-driven methodologies, supply chain operations can achieve enhanced efficiency, sustainability, and scalability. The study concludes with a discussion on current limitations, future research directions, and the potential of hybrid AI techniques for further optimizing VRP solutions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Ai-driven optimization of vehicle routing problems in supply chain: integrating graph neural networks and reinforcement learning

  • 期刊论文

  • 发表日期：

    2026年1月13日

查看全文
  • Oussama Zabraoui,
  • Anas Chafi,
  • Salaheddine Kammouri Alami

摘要

The Vehicle Routing Problem (VRP) is a critical challenge in supply chain logistics, aiming to optimize delivery routes while minimizing costs, travel time, and environmental impact. Traditional approaches, including exact algorithms, heuristics, and metaheuristics, struggle with scalability and adaptability in real-world logistics scenarios. The advent of artificial intelligence (AI) has revolutionized VRP optimization, particularly through reinforcement learning (RL) and graph neural networks (GNNs). RL formulates VRP as a sequential decision-making task, allowing adaptive and real-time route optimization. When combined with GNNs, which effectively model transportation networks as structured graphs, AI-driven methods enhance spatial awareness, optimize route selection, and improve computational efficiency. This study explores the integration of RL and GNNs in solving VRP and its key variants, including Capacitated VRP (CVRP), VRP with Time Windows (VRPTW), and Capacitated VRP with Time Windows (CVRPTW). The research presents a comparative evaluation of traditional versus AI-driven VRP solutions, analyzing performance metrics such as total travel distance, cost minimization, and computational efficiency. Experimental results demonstrate that the proposed AI framework outperforms conventional techniques by dynamically adapting to constraints and real-time disruptions. By leveraging AI-driven methodologies, supply chain operations can achieve enhanced efficiency, sustainability, and scalability. The study concludes with a discussion on current limitations, future research directions, and the potential of hybrid AI techniques for further optimizing VRP solutions.