AI-Powered Digital Twin of Urban Road Networks for Real-Time Traffic Congestion Prediction in the Metaverse

This study presents the development of an artificial intelligence-powered digital twin framework designed to predict urban traffic congestion using a Long Short-Term Memory (LSTM) deep learning model. Historical traffic data collected from multiple city road segments were analyzed to capture temporal dependencies and fluctuations in vehicular flow patterns. The proposed model was trained using a 30-day look-back window and optimized through parameter tuning, achieving a Root Mean Squared Error (RMSE) of 2,726.36 and a Mean Absolute Error (MAE) of 2,154.84. These results demonstrate the model’s capability to accurately represent complex non-linear relationships inherent in urban traffic dynamics. The temporal analysis revealed distinct bi-modal patterns corresponding to morning and evening rush hours, while a 30-day heatmap visualization highlighted recurring congestion peaks and low-traffic intervals. The integration of predictive analytics into a digital twin environment enables real-time visualization of congestion propagation, supporting data-driven planning and decision-making within a metaverse-based urban simulation. This framework establishes a methodological foundation for intelligent transportation systems that leverage artificial intelligence, digital twin technology, and virtual environments to enhance traffic forecasting, operational efficiency, and smart city management.