Digital Twins for Intelligent Intersections: A Literature Review

TL;DR

This literature review analyzes how digital twins can transform intelligent intersections by aggregating architectures, data processing, AI-driven control, VRU safety, and citywide scaling. It reveals multi-layer, interoperable DTs with real-time data fusion and AI decision-making that improve efficiency and safety, supported by co-simulation platforms and advanced VRU analytics. Yet, it also identifies critical gaps, including the simulation-reality gap, pedestrian-centric priorities, ethical governance, and validation under sparse data, calling for bidirectional, closed-loop deployments and synthetic-data–assisted learning. The work provides a roadmap for safer, more inclusive, and scalable urban mobility systems, emphasizing practical strategies such as lightweight, distributed models, robust data governance, and stakeholder-inclusive design.

Abstract

Intelligent intersections play a pivotal role in urban mobility, demanding innovative solutions such as digital twins to enhance safety and efficiency. This literature review investigates the integration and application of digital twins for intelligent intersections, a critical area within smart urban traffic systems. The review systematically categorizes existing research into five key thematic areas: (i) Digital Twin Architectures and Frameworks; (ii) Data Processing and Simulation Techniques; (iii) Artificial Intelligence and Machine Learning for Adaptive Traffic Control; (iv) Safety and Protection of Vulnerable Road Users; and (v) Scaling from Localized Intersections to Citywide Traffic Networks. Each theme is explored comprehensively, highlighting significant advancements, current challenges, and critical insights. The findings reveal that multi-layered digital twin architectures incorporating real-time data fusion and AI-driven decision-making enhances traffic efficiency and safety. Advanced simulation techniques combined with sophisticated AI/ML algorithms demonstrate notable improvements in real-time responsiveness and predictive accuracy for traffic management. Additionally, the integration of digital twins has shown substantial promise in safeguarding vulnerable road users through proactive and adaptive safety strategies. Despite these advancements, key challenges persist, including interoperability of diverse data sources, scalability of digital twins for extensive traffic networks, and managing uncertainty within dynamic urban environments. Addressing these challenges will be essential for the future development and deployment of intelligent, adaptive, and sustainable intersection management systems.

Figures (2)

• Figure 1: Number of papers by publication year.
• Figure 2: Layered digital twin architecture for intelligent intersections, shown left to right: (i) Data Acquisition Layer ingests heterogeneous real-time inputs (e.g., video, LiDAR, radar, loop detectors, weather stations); (ii) Communication Layer ensures low-latency, reliable transfer via technologies such as DSRC/WAVE, C-V2X, 5G NR, MQTT, and edge microservices; (iii) Processing Layer performs simulation, AI inference, data fusion and optimization (e.g., SUMO, GMM & EM, GNNs, generative models); (iv) Modeling Layer constructs predictive and physics-informed representations including foundation models, mesh reconstruction, and parametric assemblies; and (v) Visualization Layer supports operator interaction and asset management through web dashboards, BIM viewers, and mixed-reality interfaces. The directional flow highlights how raw sensing progresses through communication and computation into actionable models and user-facing insights.