Automated Highway Pavement Management Systems: From Inspection to Maintenance

Abstract

Highway pavement assets require intensive management due to their large scale and deteriorating nature. Traditional pavement management depends on inspections using dedicated sensors and vehicles, which are both infrequent and costly. Moreover, maintenance activities face unavoidable constraints because they rely on human labor. In this study, we explore the potential of automation in two key steps over the highway pavement management cycle: inspection and maintenance. First, we propose an automated pavement monitoring system that utilizes non-dedicated vehicle sensors, including accelerometers and dashcam cameras. Pavement condition data collected by dedicated inspection vehicles are utilized as ground truth for machine learning processes. Beyond assessing current conditions, we estimate pavement deterioration models using collected panel data. These models enable service life predictions and support optimal asset management strategies to minimize expected life cycle costs over long-term planning horizons. Second, we discuss the benefits of automated pavement maintenance technologies with AI applications, in the context of life cycle analysis.