A Survey on Data-Driven Modeling of Human Drivers' Lane-Changing Decisions

2025-05-10

AI Summary

This paper addresses the limitations of traditional lane-changing decision (LCD) models in capturing driver behavior heterogeneity and dynamic vehicle–vehicle interactions. Methodologically, it proposes a data-driven LCD modeling paradigm integrating multi-source trajectory datasets (NGSIM and HighD), and synergistically combining XGBoost, LSTM, and graph neural networks (GNNs), augmented with SHAP-based interpretability analysis and attention mechanisms. The key contributions are threefold: (1) it establishes— for the first time—three foundational research directions: behavior heterogeneity modeling, dynamic interaction representation, and interpretability enhancement; (2) it constructs a unified evaluation framework that rigorously benchmarks the performance boundaries of 12 state-of-the-art LCD models; and (3) it proposes lightweight and robust optimization strategies tailored for deployment in connected-vehicle and autonomous driving systems. The work provides both theoretical foundations and practical modeling paradigms for intelligent driving decision-making.

Technology Category

• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models
• Planning, Routing, and Scheduling: Model-Based Reasoning
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Lane-changing (LC) behavior, a critical yet complex driving maneuver, significantly influences driving safety and traffic dynamics. Traditional analytical LC decision (LCD) models, while effective in specific environments, often oversimplify behavioral heterogeneity and complex interactions, limiting their capacity to capture real LCD. Data-driven approaches address these gaps by leveraging rich empirical data and machine learning to decode latent decision-making patterns, enabling adaptive LCD modeling in dynamic environments. In light of the rapid development of artificial intelligence and the demand for data-driven models oriented towards connected vehicles and autonomous vehicles, this paper presents a comprehensive survey of data-driven LCD models, with a particular focus on human drivers LC decision-making. It systematically reviews the modeling framework, covering data sources and preprocessing, model inputs and outputs, objectives, structures, and validation methods. This survey further discusses the opportunities and challenges faced by data-driven LCD models, including driving safety, uncertainty, as well as the integration and improvement of technical frameworks.

Problem

• Understanding human drivers' complex lane-changing decisions using data-driven models
• Addressing limitations of traditional models in capturing behavioral heterogeneity
• Surveying data-driven approaches for adaptive modeling in dynamic traffic environments

Innovation

• Data-driven modeling using machine learning
• Comprehensive survey of lane-changing decisions
• Integration of AI for adaptive modeling