Leveraging tabular prior-data fitted networks for accurate pavement friction coefficient prediction from 3D texture features

Key Points

• Pavement friction coefficients can be predicted effectively using texture features, enhancing road safety.
• The study shows a prediction accuracy increase of over 20% when combining 3D texture features with fitted networks.
• Data-driven models were employed to analyze tabular prior-data networks for better forecasting of pavement friction.
• Implementing this predictive approach may lead to improved methodologies for assessing road surface effectiveness.