Integrating Graph Neural Network and GPS Data for Hourly Travel Time Estimation in Bangkok Rama-4 Road Network

This study presents a Graph Neural Network (GNN) model for accurate hourly travel time estimation in the Bangkok Rama-4 Road network. By integrating spatial and temporal data from GPS traces and OpenStreetMap geometry, the proposed model demonstrates superior performance in capturing the complex relationships between spatial and temporal information when compared to baseline methods. We evaluate the model’s performance across various scenarios, including weekdays and weekends, as well as during and outside rush hour periods. The findings of this study are expected towards significant implications for traffic management, urban planning, and delivery services, as travel time estimation is crucial for optimizing traffic flow and improving overall service quality. Future research can build upon this work by exploring additional graph neural network architectures, incorporating other relevant data sources, and investigating potential applications in various transportation sectors, to further enhance the model’s predictive capabilities and potential applications.