Artificial Intelligence and Machine Learning for Intelligent Road Construction: A Systematic Review of Pavement Materials, Construction Quality, Monitoring and Maintenance

Abstract

Road construction is increasingly becoming a data-intensive engineering activity because modern projects generate information from laboratory tests, construction equipment, global positioning systems, cameras, ground penetrating radar, thermal sensors, unmanned aerial vehicles, traffic systems and pavement-management databases. Artificial Intelligence (AI) and Machine Learning (ML) provide methods for converting these heterogeneous data into predictions, classifications and decision-support outputs. This paper presents a structured review of AI/ML applications across the road-construction and pavement-management lifecycle, including asphalt mix design, soil and subgrade assessment, intelligent paving and compaction, construction monitoring, pavement inspection, distress detection, condition prediction, deterioration forecasting and maintenance prioritization. The review synthesizes recent peer-reviewed studies, with emphasis on the engineering problem addressed, data source, model family, reported benefit and deployment limitation. The evidence shows that computer vision and deep learning are prominent in automated distress assessment, while tree-based ensembles, support vector methods and neural networks are widely used for tabular prediction. Intelligent compaction combines sensing, positioning and machine learning to improve spatial monitoring of construction quality. Recent research is also moving toward explainable AI, physics-informed neural networks and road digital twins. However, the literature remains constrained by project-specific datasets, inconsistent reporting, domain shift, limited field validation, uncertainty and weak integration with established engineering workflows. The paper argues that AI should be deployed as an engineering decision-support layer, with independent validation, uncertainty assessment and professional judgment retained for safety-critical decisions.