![[2602.21761] Survey on Neural Routing Solvers](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.21761] Survey on Neural Routing Solvers
Summary
This survey reviews Neural Routing Solvers (NRSs) that utilize deep learning for vehicle routing problems, highlighting their heuristic nature and proposing a new evaluation pipeline.
Why It Matters
As industries increasingly rely on efficient routing solutions, understanding the advancements in Neural Routing Solvers is crucial. This survey not only categorizes existing methods but also identifies gaps in research, guiding future developments in this area.
Key Takeaways
- Neural Routing Solvers leverage deep learning to improve vehicle routing efficiency.
- The survey introduces a hierarchical taxonomy based on heuristic principles.
- A new evaluation pipeline is proposed to address existing research limitations.
- Comparative benchmarking reveals gaps in current NRS research.
- Understanding NRSs can reduce reliance on manual design in routing solutions.
Mathematics > Optimization and Control arXiv:2602.21761 (math) [Submitted on 25 Feb 2026] Title:Survey on Neural Routing Solvers Authors:Yunpeng Ba, Xi Lin, Changliang Zhou, Ruihao Zheng, Zhenkun Wang, Xinyan Liang, Zhichao Lu, Jianyong Sun, Yuhua Qian, Qingfu Zhang View a PDF of the paper titled Survey on Neural Routing Solvers, by Yunpeng Ba and 9 other authors View PDF HTML (experimental) Abstract:Neural routing solvers (NRSs) that leverage deep learning to tackle vehicle routing problems have demonstrated notable potential for practical applications. By learning implicit heuristic rules from data, NRSs replace the handcrafted counterparts in classic heuristic frameworks, thereby reducing reliance on costly manual design and trial-and-error adjustments. This survey makes two main contributions: (1) The heuristic nature of NRSs is highlighted, and existing NRSs are reviewed from the perspective of heuristics. A hierarchical taxonomy based on heuristic principles is further introduced. (2) A generalization-focused evaluation pipeline is proposed to address limitations of the conventional pipeline. Comparative benchmarking of representative NRSs across both pipelines uncovers a series of previously unreported gaps in current research. Subjects: Optimization and Control (math.OC); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE) Cite as: arXiv:2602.21761 [math.OC] (or arXiv:2602.21761v1 [math.OC] for this version) http...
