Neural Combinatorial Optimization for Real-World Routing

2025-03-20

AI Summary

This paper addresses the poor generalization of Neural Combinatorial Optimization (NCO) methods in real-world logistics scenarios—where existing approaches, predominantly trained on symmetric Euclidean distances, fail to capture asymmetric road networks, realistic travel times, and complex data distributions. To bridge this gap, we propose an end-to-end Graph Neural Network (GNN) framework. Our key contributions are: (1) the Angle-Aware Attention-Free Module (AAFM), the first to jointly incorporate angular relationships and adaptive bias into attention-free node representation learning; (2) a context-gated mechanism that jointly encodes node and edge features; and (3) the first large-scale, open-source Vehicle Routing Problem (VRP) benchmark—comprising 100 cities—with real-world distance and travel-time matrices derived from the Open Source Routing Machine (OSRM). Experiments demonstrate state-of-the-art performance on realistic VRP benchmarks. The code and dataset are publicly released.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Search and Optimization: Combinatorial Optimization
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Vehicle Routing Problems (VRPs) are a class of NP-hard problems ubiquitous in several real-world logistics scenarios that pose significant challenges for optimization. Neural Combinatorial Optimization (NCO) has emerged as a promising alternative to classical approaches, as it can learn fast heuristics to solve VRPs. However, most research works in NCO for VRPs focus on simplified settings, which do not account for asymmetric distances and travel durations that cannot be derived by simple Euclidean distances and unrealistic data distributions, hindering real-world deployment. This work introduces RRNCO (Real Routing NCO) to bridge the gap of NCO between synthetic and real-world VRPs in the critical aspects of both data and modeling. First, we introduce a new, openly available dataset with real-world data containing a diverse dataset of locations, distances, and duration matrices from 100 cities, considering realistic settings with actual routing distances and durations obtained from Open Source Routing Machine (OSRM). Second, we propose a novel approach that efficiently processes both node and edge features through contextual gating, enabling the construction of more informed node embedding, and we finally incorporate an Adaptation Attention Free Module (AAFM) with neural adaptive bias mechanisms that effectively integrates not only distance matrices but also angular relationships between nodes, allowing our model to capture rich structural information. RRNCO achieves state-of-the-art results in real-world VRPs among NCO methods. We make our dataset and code publicly available at https://github.com/ai4co/real-routing-nco.

Problem

• Addresses real-world Vehicle Routing Problems (VRPs) with asymmetric distances and durations.
• Introduces RRNCO to bridge the gap between synthetic and real-world VRPs.
• Proposes a novel approach with contextual gating and neural adaptive bias mechanisms.

Innovation

• Introduces RRNCO for real-world VRPs
• Uses contextual gating for node and edge features
• Incorporates AAFM with neural adaptive bias