Developing a Pavement Management System using data fusion of physical features

Pavement assessment is still plagued by the expensive and time-consuming distress data collection method. The VOTERS (Versatile Onboard Traffic Embedded Roaming Sensors) project has successfully transformed the way pavement inspections are performed, from mainly visual distresses identification to measurements of physical pavement condition features through the use of various sensing systems mounted on a vehicle. It allows network-wide determination and pavement condition ranking of our vast highway network. This thesis presents how these physical features can substitute for traditional distress assessment methods. Then these substitute features are used as the main inputs to a new kind of Pavement Management System. The first step is to thoroughly understand the current state of practices, especially the PAVER system (MicroPaver(TM)). Next we focus on using these features, determined from sensor data, in developing a deduct value approach based on nonlinear regression for Pavement Condition Index. A decision tree based algorithm for maintenance, repair and priority programming based optimization algorithms is developed. Using data driven life cycle costs we have shown the benefits of using the VOTERS system. The primary contribution of the thesis is to develop a Pavement Management System that would be automated and that would show the economic benefits for a roadway sector that could potentially use VOTERS sensor and technology.