Machine Learning Approaches for Detection/Classification and Prediction Purposes in Pavement Engineering Studies: An Overview

Abstract

In order to maintain, manage, and budget for pavement infrastructure, road pavement condition assessment is necessary. Several pavement characteristics are measured to assess its condition, including pavement strength, roughness, and surface distresses. It is important to categorize studies at deeper levels due to the rapid growth of articles published in this field. The objective of this paper is to provide an overview of machine learning-based pavement evaluation studies and their contributions to the area. In order to facilitate the exploration of the studies employing similar methodologies, the studies are organized based on their goals. Therefore, studies are classified based on the two main categories of goals employed in them, namely: 1. Studies with aim of pavement condition prediction and 2. Studies with the aim of pavement distress detection/classification. It is observed that research of category 1 has grown very well during the past years. Also, category 2 includes studies that mostly focus on crack detection and it can be felt that there is a need for expanding the focus of studies on other types of distresses.