Machine Learning Surrogates for Optimizing Transportation Policies with Agent-Based Models

2025-01-19

AI Summary

To address traffic congestion and air pollution exacerbated by urban population growth, this paper proposes a novel graph neural network (GNN)-based paradigm for urban transportation policy evaluation. Moving beyond computationally expensive agent-based simulation models (e.g., MATSim), we introduce the first interpretable GNN surrogate model that jointly captures macro-level policy response dynamics and micro-level road network topology. Our model operates at the edge level, explicitly encoding policy sensitivity features to enhance both generalizability and interpretability. Empirical evaluation on Paris demonstrates an average absolute error of less than 8.2% in predicting traffic flow changes on policy-affected and high-volume road segments. The model enables sub-second evaluation across hundreds of policy scenarios—accelerating analysis by over two orders of magnitude compared to MATSim—while preserving transparency and structural fidelity.

Technology Category

• Planning, Routing, and Scheduling: Model-Based Reasoning
• Machine Learning: Graph-based Machine Learning
• Search and Optimization: Sampling/Simulation-based Search

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 evaluation methodologies and metrics
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Rapid urbanization and growing urban populations worldwide present significant challenges for cities, including increased traffic congestion and air pollution. Effective strategies are needed to manage traffic volumes and reduce emissions. In practice, traditional traffic flow simulations are used to test those strategies. However, high computational intensity usually limits their applicability in investigating a magnitude of different scenarios to evaluate best policies. This paper presents a first approach of using Graph Neural Networks (GNN) as surrogates for large-scale agent-based simulation models. In a case study using the MATSim model of Paris, the GNN effectively learned the impacts of capacity reduction policies on citywide traffic flow. Performance analysis across various road types and scenarios revealed that the GNN could accurately capture policy-induced effects on edge-based traffic volumes, particularly on roads directly affected by the policies and those with higher traffic volumes.

Problem

• Urban Traffic Control
• Pollution Reduction
• Population Growth

Innovation

• Graph Neural Networks
• Traffic Simulation
• Road Capacity Prediction