![[2510.16701] An Agentic Framework with LLMs for Solving Complex Vehicle Routing Problems](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2510.16701] An Agentic Framework with LLMs for Solving Complex Vehicle Routing Problems
Summary
The paper presents an Agentic Framework with Large Language Models (LLMs) designed to automate complex vehicle routing problems (VRPs), enhancing solution feasibility and code reliability.
Why It Matters
This research addresses the persistent challenges in vehicle routing, a critical area in logistics and transportation. By leveraging LLMs for full automation, it promises to reduce reliance on human intervention, potentially transforming operational efficiencies in various industries.
Key Takeaways
- The Agentic Framework (AFL) automates VRP solutions from input to output.
- AFL improves trustworthiness by decomposing tasks and using specialized agents.
- The framework outperforms existing LLM-based methods in code reliability and solution feasibility.
Computer Science > Artificial Intelligence arXiv:2510.16701 (cs) [Submitted on 19 Oct 2025 (v1), last revised 14 Feb 2026 (this version, v2)] Title:An Agentic Framework with LLMs for Solving Complex Vehicle Routing Problems Authors:Ni Zhang, Zhiguang Cao, Jianan Zhou, Cong Zhang, Yew-Soon Ong View a PDF of the paper titled An Agentic Framework with LLMs for Solving Complex Vehicle Routing Problems, by Ni Zhang and 4 other authors View PDF HTML (experimental) Abstract:Complex vehicle routing problems (VRPs) remain a fundamental challenge, demanding substantial expert effort for intent interpretation and algorithm design. While large language models (LLMs) offer a promising path toward automation, current approaches still rely on external intervention, which restrict autonomy and often lead to execution errors and low solution feasibility. To address these challenges, we propose an Agentic Framework with LLMs (AFL) for solving complex vehicle routing problems, achieving full automation from problem instance to solution. AFL directly extracts knowledge from raw inputs and enables self-contained code generation without handcrafted modules or external solvers. To improve trustworthiness, AFL decomposes the overall pipeline into three manageable subtasks and employs four specialized agents whose coordinated interactions enforce cross-functional consistency and logical soundness. Extensive experiments on 60 complex VRPs, ranging from standard benchmarks to practical variants, val...
