# Learn to Design the Heuristics for Vehicle Routing Problem

arXiv: https://arxiv.org/abs/2002.08539 · HTML: https://mlsanity.com/paper/2002.08539/

## Key facts

- Cited by 24 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Feb 2020.

## Built on by (most cited first, 24 of 24)

- [Deep Policy Dynamic Programming for Vehicle Routing Problems](https://mlsanity.com/paper/2102.11756/) (2021)
- [Learning Large Neighborhood Search Policy for Integer Programming](https://mlsanity.com/paper/2111.03466/) (2021)
- [Cross-Problem Learning for Solving Vehicle Routing Problems](https://mlsanity.com/paper/2404.11677/) (2024)
- [CaDA: Cross-Problem Routing Solver with Constraint-Aware Dual-Attention](https://mlsanity.com/paper/2412.00346/) (2024)
- [Analytics and Machine Learning in Vehicle Routing Research](https://mlsanity.com/paper/2102.10012/) (2021)
- [An Overview and Experimental Study of Learning-based Optimization Algorithms for Vehicle Routing Problem](https://mlsanity.com/paper/2107.07076/) (2021)
- [Learning to Solve Vehicle Routing Problems: A Survey](https://mlsanity.com/paper/2205.02453/) (2022)
- [Solve routing problems with a residual edge-graph attention neural network](https://mlsanity.com/paper/2105.02730/) (2021)
- [Learning Large Neighborhood Search for Vehicle Routing in Airport Ground Handling](https://mlsanity.com/paper/2302.13797/) (2023)
- [Learning to Solve Multiple-TSP with Time Window and Rejections via Deep Reinforcement Learning](https://mlsanity.com/paper/2209.06094/) (2022)
- [Online Control of Adaptive Large Neighborhood Search using Deep Reinforcement Learning](https://mlsanity.com/paper/2211.00759/) (2022)
- [A machine learning framework for neighbor generation in metaheuristic search](https://mlsanity.com/paper/2212.11451/) (2022)
- [Survey on Neural Routing Solvers](https://mlsanity.com/paper/2602.21761/) (2026)
- [The First AI4TSP Competition: Learning to Solve Stochastic Routing Problems](https://mlsanity.com/paper/2201.10453/) (2022)
- [Route Planning for Last-Mile Deliveries Using Mobile Parcel Lockers: A Hybrid Q-Learning Network Approach](https://mlsanity.com/paper/2209.04265/) (2022)
- [A review of approaches to modeling applied vehicle routing problems](https://mlsanity.com/paper/2105.10950/) (2021)
- [Learning Enhanced Optimisation for Routing Problems](https://mlsanity.com/paper/2109.08345/) (2021)
- [Multi-objective Pointer Network for Combinatorial Optimization](https://mlsanity.com/paper/2204.11860/) (2022)
- [Location-routing Optimisation for Urban Logistics Using Mobile Parcel Locker Based on Hybrid Q-Learning Algorithm](https://mlsanity.com/paper/2110.15485/) (2021)
- [Combining Constructive and Perturbative Deep Learning Algorithms for the Capacitated Vehicle Routing Problem](https://mlsanity.com/paper/2211.13922/) (2022)
- [A Coalition Game for On-demand Multi-modal 3D Automated Delivery System](https://mlsanity.com/paper/2412.17252/) (2024)
- [Learning to Search for Vehicle Routing with Multiple Time Windows](https://mlsanity.com/paper/2505.23098/) (2025)
- [LRM-1B: Towards Large Routing Model](https://mlsanity.com/paper/2507.03300/) (2025)
- [Constraints Matrix Diffusion based Generative Neural Solver for Vehicle Routing Problems](https://mlsanity.com/paper/2603.07568/) (2026)

## Abstract (arXiv)

This paper presents an approach to learn the local-search heuristics that iteratively improves the solution of Vehicle Routing Problem (VRP). A local-search heuristics is composed of a destroy operator that destructs a candidate solution, and a following repair operator that rebuilds the destructed one into a new one. The proposed neural network, as trained through actor-critic framework, consists of an encoder in form of a modified version of Graph Attention Network where node embeddings and edge embeddings are integrated, and a GRU-based decoder rendering a pair of destroy and repair operators. Experiment results show that it outperforms both the traditional heuristics algorithms and the existing neural combinatorial optimization for VRP on medium-scale data set, and is able to tackle the large-scale data set (e.g., over 400 nodes) which is a considerable challenge in this area. Moreover, the need for expertise and handcrafted heuristics design is eliminated due to the fact that the proposed network learns to design the heuristics with a better performance. Our implementation is available online.