Assess and refine data quality and data management issues using 3D technology and automation for optimized pavement management
Steele, Ariel Nicole
MetadataShow full item record
Despite the growing popularity of 3D automated pavement data collection systems, there has been very little research on the inherent variability of pavement distress data and no existing research on the underlying causes of differences between multiple runs of 3D data collection. Thus, there is a need to quantify the inherent variability of repeated crack measurements, to determine the factors that impact the quality and consistency of crack detection results, and to identify steps that may be taken to systematically improve the reliability of automatic pavement condition data. The methodology of this study is divided into two stages using different datasets to achieve the research objectives. First, a crack variability study is conducted to observe differences in data collection between runs, both qualitatively and quantitatively. Second, a crack deterioration study using historical interstate data is done to investigate potential data quality issues for pavement forecasting and specifically to quantify the effects of poor data registration. The key contributions of this study are the following: 1) developed a methodology for quantifying crack variability between multiple runs of 3D pavement data collection, 2) quantified the inherent variability of crack detection between repeated runs of data collection, 3) determined some of the reasons for the changes in crack detection results between runs of data collection, 4) identified potential data quality issues in historical 3D pavement data, and 5) provided recommendations for transportation agencies and data vendors to ensure high-quality and reliable pavement condition results.