Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Large Language Models and Their Applications in Roadway Safety and Mobility Enhancement: A Comprehensive Review

Muhammad Monjurul Karim, Yan Shi, Shucheng Zhang, Bingzhang Wang, Mehrdad Nasri, Yinhai Wang

TL;DR

This paper addresses the challenge of improving roadway safety and mobility by surveying how Large Language Models (LLMs) can be adapted for transportation. It systematically classifies foundational concepts, architectural innovations, and prompting strategies (including RAG and LLM-as-orchestrator) and reviews a broad suite of mobility and safety applications, emphasizing multimodal integration and edge-enabled deployment. The authors highlight enabling technologies such as V2X, foundation models, explainability frameworks, and safety verification, while candidly discussing persistent barriers like hallucinations, grounding in physical reality, data governance, and sim-to-real gaps. They also outline a roadmap of future research directions—advanced multimodal fusion, native spatio-temporal reasoning, causal inference, human-AI collaboration, continuous learning, and domain-specific efficient models—to realize practical, trustworthy LLM-powered transportation systems with real-world impact.

Abstract

Roadway safety and mobility remain critical challenges for modern transportation systems, demanding innovative analytical frameworks capable of addressing complex, dynamic, and heterogeneous environments. While traditional engineering methods have made progress, the complexity and dynamism of real-world traffic necessitate more advanced analytical frameworks. Large Language Models (LLMs), with their unprecedented capabilities in natural language understanding, knowledge integration, and reasoning, represent a promising paradigm shift. This paper comprehensively reviews the application and customization of LLMs for enhancing roadway safety and mobility. A key focus is how LLMs are adapted -- via architectural, training, prompting, and multimodal strategies -- to bridge the "modality gap" with transportation's unique spatio-temporal and physical data. The review systematically analyzes diverse LLM applications in mobility (e.g., traffic flow prediction, signal control) and safety (e.g., crash analysis, driver behavior assessment,). Enabling technologies such as V2X integration, domain-specific foundation models, explainability frameworks, and edge computing are also examined. Despite significant potential, challenges persist regarding inherent LLM limitations (hallucinations, reasoning deficits), data governance (privacy, bias), deployment complexities (sim-to-real, latency), and rigorous safety assurance. Promising future research directions are highlighted, including advanced multimodal fusion, enhanced spatio-temporal reasoning, human-AI collaboration, continuous learning, and the development of efficient, verifiable systems. This review provides a structured roadmap of current capabilities, limitations, and opportunities, underscoring LLMs' transformative potential while emphasizing the need for responsible innovation to realize safer, more intelligent transportation systems.

Large Language Models and Their Applications in Roadway Safety and Mobility Enhancement: A Comprehensive Review

TL;DR

This paper addresses the challenge of improving roadway safety and mobility by surveying how Large Language Models (LLMs) can be adapted for transportation. It systematically classifies foundational concepts, architectural innovations, and prompting strategies (including RAG and LLM-as-orchestrator) and reviews a broad suite of mobility and safety applications, emphasizing multimodal integration and edge-enabled deployment. The authors highlight enabling technologies such as V2X, foundation models, explainability frameworks, and safety verification, while candidly discussing persistent barriers like hallucinations, grounding in physical reality, data governance, and sim-to-real gaps. They also outline a roadmap of future research directions—advanced multimodal fusion, native spatio-temporal reasoning, causal inference, human-AI collaboration, continuous learning, and domain-specific efficient models—to realize practical, trustworthy LLM-powered transportation systems with real-world impact.

Abstract

Roadway safety and mobility remain critical challenges for modern transportation systems, demanding innovative analytical frameworks capable of addressing complex, dynamic, and heterogeneous environments. While traditional engineering methods have made progress, the complexity and dynamism of real-world traffic necessitate more advanced analytical frameworks. Large Language Models (LLMs), with their unprecedented capabilities in natural language understanding, knowledge integration, and reasoning, represent a promising paradigm shift. This paper comprehensively reviews the application and customization of LLMs for enhancing roadway safety and mobility. A key focus is how LLMs are adapted -- via architectural, training, prompting, and multimodal strategies -- to bridge the "modality gap" with transportation's unique spatio-temporal and physical data. The review systematically analyzes diverse LLM applications in mobility (e.g., traffic flow prediction, signal control) and safety (e.g., crash analysis, driver behavior assessment,). Enabling technologies such as V2X integration, domain-specific foundation models, explainability frameworks, and edge computing are also examined. Despite significant potential, challenges persist regarding inherent LLM limitations (hallucinations, reasoning deficits), data governance (privacy, bias), deployment complexities (sim-to-real, latency), and rigorous safety assurance. Promising future research directions are highlighted, including advanced multimodal fusion, enhanced spatio-temporal reasoning, human-AI collaboration, continuous learning, and the development of efficient, verifiable systems. This review provides a structured roadmap of current capabilities, limitations, and opportunities, underscoring LLMs' transformative potential while emphasizing the need for responsible innovation to realize safer, more intelligent transportation systems.

Paper Structure

This paper contains 68 sections, 7 equations, 2 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related works
  3. Foundational Concepts: LLMs in the Transportation Context
  4. Core LLM Concepts in the transportation context
  5. LLM training, fine-tuning, and domain adaptation:
  6. Prompt Engineering
  7. Retrieval-Augmented Generation (RAG):
  8. LLM Agents and tool use
  9. Multimodal Large Language Models
  10. Specialized Architectures for Transportation Applications
  11. Integration of Explicit Spatio-Temporal Modules
  12. Novel Input Representation and Tokenization Strategies
  13. Partially Frozen Model Adaptation
  14. LLM-as-Orchestrator Integration Frameworks
  15. LLM Applications for Mobility Enhancement
  16. ...and 53 more sections

Figures (2)

  • Figure 1: Conceptual framework for augmenting Large Language Models (LLMs) with explicit Spatio-Temporal (S-T) modules for transportation applications.
  • Figure 2: Overview of input representation strategies for transforming spatio-temporal transportation data into LLM-compatible formats.