Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
84 Pavement condition data have been collected, analyzed, and used by highway agencies for decades. Data collection has progressed from labor-intensive manual pavement condition surveys conducted using pen and paper to collection of 3D images for 100% of the surveyed lanes at posted highway speeds. Data analysis has also progressed. With the advent of 3D imaging, analyses have progressed to automated distress detection algorithms that require minimal to no user intervention. As noted in the FAST Act, data quality has become an increasingly important criteria with automated pavement condition data collection. This is due not only to the criticality of decisions based on the pavement condition data, but also to the speed at which the data are collected. Since DCVs are capable of collecting data at highway speeds, it is imperative that data quality be checked and maintained as data collection progresses rather than waiting to check quality after data collection is complete. Overall Findings Automated imaging technology includes 2D- and 3D-laser scanning and capturing digital images. 2D imaging was first introduced in 2006 and was able to produce clear images; however, there were significant challenges in producing fully automated distress identification. Several years later, 3D imaging from the manufacturing industry was applied to assessing the condition of the pavement surface. Through the use of downward-facing cameras and infrared lasers, it was easier to use computer programs to automatically detect and quantify pavement surface distress. Collection of data and images at posted highway speeds quickly results in a need for management of raw and processed data. Depending on the type and amount of data and images collected, data storage requirements can easily exceed 1 GB per mile. Data quality management plans assist agencies in documenting, tracking, and evaluating the acceptability of data and images collected during the pavement condition survey. Data quality management plans include processes for QC, acceptance, reporting, and corrective action. The majority of agencies are storing images, raw data, calculated condition indices, and distress data. Data are stored in spreadsheet or database formats, and images are stored in JPG format. Three agencies reported accessing data and images via a vendor-hosted site. In relation to data and image retention, four agencies retain both the data and the images indefinitely, two agencies retain only the data indefinitely, and three agencies indicated a retention schedule of 4 to 10 years. C H A P T E R 6 Conclusions
Conclusions 85 Direct comparison of agency cost for pavement condition surveys is challenging, for example, due to network size, duration of data collection, extent of distress types collected, semi- versus fully automated analysis, agency- versus vendor-based costs, etc. One agency with a relatively small network that requires the vendor to conduct fully automated surveys reported a cost of $43/mi ($27/km). For semiautomated analysis, costs ranged from $34 to $101/mi ($21 to $63/km), and when both semi- and fully automated methods are required, costs ranged from $28 to $115/mi ($17 to $71/km). Based on the agency survey, approximately 90% of the responding agencies indicated using automated pavement condition surveys. Some of the noted successes of automated condition surveys include, for example, more efficient, safer, and faster data collection compared to manual surveys; more consistent data; time savings with fully automated crack detection; and success in extracting asset data from ROW images. Agencies also indicated a number of challenges with automated data collection, for example, establishing ground truth for con- firming automated results, standardizing methods for quantifying distress, integrating LRS into the pavement management system, and quantifying the more challenging distress types (e.g., patching, raveling, bleeding). Three case examples were prepared summarizing the data quality approach for the British Columbia MoTI, Pennsylvania DOT, and North Dakota DOT. These three case examples showcase vendor versus agency data collection and analysis and semi- versus fully automated data collection. Suggestions for Future Research The following bullet points present research suggestions from agencies during the follow-up questions. â¢ Develop a standardized method for evaluating automated pavement distress algorithms. A number of algorithms have been developed to detect and classify distress type and severity. However, there is no standardized method for evaluating the performance (and accuracy) of these algorithms and methods. This research would develop a method to evaluate the applicability and accuracy of algorithms used to identify pavement distress type and severity. â¢ Improve the accuracy of automated crack detection on high macrotexture surfaces. Depending on aggregate size, chip-sealed surfaced pavements, for example, can have a relatively high surface macrotexture. This level of macrotexture can impact the ability of automated crack detection algorithms to identify cracking high macrotexture surfaces. This research would determine the impact of macrotexture on automated crack detection algo- rithms and software and develop a method for accurately determining crack type and severity on high macrotexture surfaces. â¢ Determine the effort needed to establish state or regional DCV certification facilities. Similar to falling weight deflectometer calibration centers, this research would determine and recommend, for example, initial setup and operating cost, test procedures, requirements (equipment, hardware and software, analysis methods), and analysis process for certify- ing DCVs and data collection personnel within a given agency or at a regional calibration center. â¢ Fully validate the methods for determining IRI values from 3D profile measurements. Additional research is needed to fully validate the accuracy and repeatability of methods for determining IRI values from 3D profile measurements. This effort should also include development of a standard, for AASHTO consideration, for determining IRI from 3D profile measurements.
86 Automated Pavement Condition Surveys â¢ Study the impact of changing pavement condition data collection equipment or service provider. Highway agencies conduct automated pavement condition surveys using agency- owned DCVs, through service provider contracts, or a combination of both. Guidance on the impact of transitioning from one system to the next is needed in the event that different equipment is purchased or a different service provider is selected. This guidance, for example, could potentially include methods for evaluating new equipment prior to purchase or service provider evaluation during the proposal process, criteria for determining data acceptability, analysis methods for comparing results, etc.