DPN: Decoupling Partition and Navigation for Neural Solvers of Min-max Vehicle Routing Problems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zheng, Zhi, Yao, Shunyu, Wang, Zhenkun, Tong, Xialiang, Yuan, Mingxuan, Tang, Ke
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917686048456704
author Zheng, Zhi
Yao, Shunyu
Wang, Zhenkun
Tong, Xialiang
Yuan, Mingxuan
Tang, Ke
author_facet Zheng, Zhi
Yao, Shunyu
Wang, Zhenkun
Tong, Xialiang
Yuan, Mingxuan
Tang, Ke
contents The min-max vehicle routing problem (min-max VRP) traverses all given customers by assigning several routes and aims to minimize the length of the longest route. Recently, reinforcement learning (RL)-based sequential planning methods have exhibited advantages in solving efficiency and optimality. However, these methods fail to exploit the problem-specific properties in learning representations, resulting in less effective features for decoding optimal routes. This paper considers the sequential planning process of min-max VRPs as two coupled optimization tasks: customer partition for different routes and customer navigation in each route (i.e., partition and navigation). To effectively process min-max VRP instances, we present a novel attention-based Partition-and-Navigation encoder (P&N Encoder) that learns distinct embeddings for partition and navigation. Furthermore, we utilize an inherent symmetry in decoding routes and develop an effective agent-permutation-symmetric (APS) loss function. Experimental results demonstrate that the proposed Decoupling-Partition-Navigation (DPN) method significantly surpasses existing learning-based methods in both single-depot and multi-depot min-max VRPs. Our code is available at
format Preprint
id arxiv_https___arxiv_org_abs_2405_17272
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle DPN: Decoupling Partition and Navigation for Neural Solvers of Min-max Vehicle Routing Problems
Zheng, Zhi
Yao, Shunyu
Wang, Zhenkun
Tong, Xialiang
Yuan, Mingxuan
Tang, Ke
Machine Learning
Artificial Intelligence
The min-max vehicle routing problem (min-max VRP) traverses all given customers by assigning several routes and aims to minimize the length of the longest route. Recently, reinforcement learning (RL)-based sequential planning methods have exhibited advantages in solving efficiency and optimality. However, these methods fail to exploit the problem-specific properties in learning representations, resulting in less effective features for decoding optimal routes. This paper considers the sequential planning process of min-max VRPs as two coupled optimization tasks: customer partition for different routes and customer navigation in each route (i.e., partition and navigation). To effectively process min-max VRP instances, we present a novel attention-based Partition-and-Navigation encoder (P&N Encoder) that learns distinct embeddings for partition and navigation. Furthermore, we utilize an inherent symmetry in decoding routes and develop an effective agent-permutation-symmetric (APS) loss function. Experimental results demonstrate that the proposed Decoupling-Partition-Navigation (DPN) method significantly surpasses existing learning-based methods in both single-depot and multi-depot min-max VRPs. Our code is available at
title DPN: Decoupling Partition and Navigation for Neural Solvers of Min-max Vehicle Routing Problems
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2405.17272