Traffic Digital Twin

Definition and Technical Concept

Traffic Digital Twin refers to a virtual representation of real-world traffic systems built using digital twin technology. It gathers real-world data across transport infrastructure, road users (motorised vehicles, non-motorised transport, and pedestrians), traffic movements (origins, destinations, and routes), and surrounding environments (weather conditions, urban functional zones, etc.). Its core concept lies in establishing a "digital twin" that precisely mirrors and synchronises with physical traffic systems in real time, empowering transport authorities to "gain insight into and simulate physical-world traffic operations within a virtual environment". Beyond simply "monitoring" live road conditions, its primary role is to drive informed decision-making through simulation, predictive modelling, and optimisation.

Core Architecture and Technical Framework

A comprehensive traffic digital twin system typically adopts a four-tier architecture spanning perception to decision-making. The Domain-Wide Perception Access Layer serves as the primary data intake, establishing a unified spatial data gateway via multi-source sensors such as roadside CCTV surveillance, millimetre-wave (mmWave) radar, geomagnetic sensors, checkpoint systems, and vehicle-to-everything (V2X) terminals. The Spatio-Temporal Engine Foundation Layer acts as the system's core computing engine, responsible for mapping 2D perception data onto a 3D spatial coordinate system to execute real-time spatial binding and computing between dynamic targets (such as vehicles and pedestrians) and road geometries. The Embodied Cloud Intelligence Layer functions as the central decision-making hub, converting raw inputs into actionable traffic insights via AI models and simulation engines. The Industry Decision Application Layer caters directly to traffic management authorities, delivering practical applications such as traffic light signal optimisation, congestion alerts, and emergency response dispatch.

Key Technologies and Evolutionary Trends

The development of traffic digital twins relies on several critical technologies. In terms of data fusion, integrating multi-source inputs—including roadside CCTV feeds, floating car GPS data, and radar—under a unified spatio-temporal benchmark resolves longstanding pain points where objects are "visible but cannot be located, or identifiable but impossible to track continuously". For model development, the system must incorporate macroscopic traffic flow models (to evaluate city-wide phenomena like congestion spread) alongside microscopic traffic flow models (to simulate individual vehicle behaviour and interaction), thereby achieving multi-scale digital mapping from macro to micro levels. Architecturally, an edge-cloud collaborative framework offloads real-time perception processing to roadside edge nodes, transmitting only streamlined structured data to the cloud for aggregation to cut bandwidth overhead and maintain rapid system response times.

Industry Deployment and Practical Outcomes

Traffic digital twin solutions have already been deployed across numerous local and international scenarios. The Fujian Xiamen North Road Network Management Sub-centre deployed a digital twin management platform powered by high-precision 3D modelling, combining mmWave radar and dual telephoto/short-focus cameras to transition from "reactive response" to "proactive intelligent traffic management", earning recognition as a national benchmark case in highway tunnel digitalisation. In Fresno, California, the SMART-TWIN project leverages digital twin infrastructure alongside AI-driven microscopic traffic simulation and machine learning classifiers to execute real-time detection and impact assessments for scenarios such as signal failures and lane closures, supporting active traffic management strategies. In academia, frameworks like FlowTwin have completed city-scale validation in Bologna, Italy, achieving high-fidelity traffic flow simulation alongside real-time monitoring.

Technical Limitations

The operational efficacy of traffic digital twins depends heavily on the sensor coverage density and data quality of front-end devices, real-time multi-source data fusion capabilities, and the stability of communication networks. As the technology transitions from pure "visualisation" to "predictive, closed-loop execution", existing systems still face significant engineering hurdles, including real-time synchronised computation across massive road networks, cross-system interoperability, and model generalisation. Moreover, high upfront deployment costs and ongoing maintenance expenditure remain tangible barriers to widespread market adoption.