RouteFinder: Towards Foundation Models for Vehicle Routing Problems

TL;DR

RouteFinder is a transformer-based foundation model framework designed to handle various Vehicle Routing Problem variants efficiently, using attribute embeddings, reinforcement learning techniques, and adapter layers for fine-tuning.

Contribution

The paper introduces a unified VRP environment and a transformer-based foundation model that can adapt to multiple VRP variants with unseen attributes.

Findings

Outperforms recent state-of-the-art learning methods on 48 VRP variants.

Uses reinforcement learning techniques like mixed batch training and reward normalization.

Employs adapter layers for efficient fine-tuning on new VRP variants.

Abstract

This paper introduces RouteFinder, a comprehensive foundation model framework to tackle different Vehicle Routing Problem (VRP) variants. Our core idea is that a foundation model for VRPs should be able to represent variants by treating each as a subset of a generalized problem equipped with different attributes. We propose a unified VRP environment capable of efficiently handling any combination of these attributes. The RouteFinder model leverages a modern transformer-based encoder and global attribute embeddings to improve task representation. Additionally, we introduce two reinforcement learning techniques to enhance multi-task performance: mixed batch training, which enables training on different variants at once, and multi-variant reward normalization to balance different reward scales. Finally, we propose efficient adapter layers that enable fine-tuning for new variants with unseen attributes. Extensive experiments on 48 VRP variants show RouteFinder outperforms recent state-of-the-art learning methods. Our code is publicly available at https://github.com/ai4co/routefinder.