Abstract Manager

Pothole-induced damage to both road users and vehicles has been recently increasing at an alarming rate, posing significant challenges to road integrity and safety. Routine pothole detection, repair, and maintenance is a critical task to sustain and preserve transportation infrastructure. This study introduces a novel method for pothole detection and quantification using low-cost, low-point-density LiDAR sensors to address the need for cost-effective road maintenance solutions. Traditional methods, although varied, often lack precision, depth analysis, and cost-efficiency, especially for large-scale network assessment. Existing image-based methods rely on strict image quality conditions and require large computational resources. High-resolution LiDAR, while accurate, poses cost and computational challenges. In this research, we proposed a method that attempts to overcome existing challenges by using low-cost LiDAR data to detect potholes. First, surface curvature is estimated for each point in the LiDAR-generated point cloud, and local surface fitting and eigenvalue computation from covariance matrices are obtained. Based on these curvature estimates, the point cloud is segmented to identify potential potholes. The pothole boundaries are delineated using convex hull operations, and the quantification process computes the dimensions, area, and volume of each pothole. A comparative analysis is conducted to examine the balance between pothole detection and quantification accuracy and LiDAR point density. The findings indicate that LiDAR sensors, costing up to eight times less and collecting point cloud datasets with sixteen times lower point density than high-end sensors, can effectively detect potholes with adequate accuracy. The compromise in terms of pothole quantification, spatial resolution, and detection precision is minimal.  This research aims to provide a scalable, economical solution for comprehensive road network assessments for transportation entities interested in low-entry data acquisition systems.