Pavement asset management systems and technologies: A review

Highlights

• Documented chronological developments in Pavement Asset Management Systems

• Reviewed pavement condition state-of-the-art data collection technologies

• Collected image processing and machine learning applications in PAMS data analysis

• Discussed utilization of various heuristic techniques in effective decision making

• Collation of literature indicative of vast scope to use AI in automating PAMS

Abstract

Pavement asset management system (PAMS) assists agencies and decision makers to maintain deteriorating pavement assets with optimized budget allocation. The recent developments in pavement condition data collection and processing have significant effect on estimating remaining service life and selecting optimum maintenance strategies. Further, image processing (IP) and artificial intelligence (AI) tools have improved the overall performance of PAMS by helping analyze big data emanating from distress surveys. The objective of this review paper was to collect and report several current state-of-the-art developments in PAMS and the associated embedded processes, majorly focused on data collection procedures, analytical techniques, decision making tools, and processing methods. The shift from manual condition surveys to automated pavement condition surveys has profusely improved data collection rate. The wide-range of data collection methods, manual, automated vehicles, and cost-effective methods followed across the globe were reviewed. Further, the chronological development in data analysis, specifically, distress evaluation, homogeneous sectioning for selection of maintenance strategies, and prioritization and optimization of maintenance strategies were discussed while emphasizing the application of IP and AI in enhancing the efficacy of PAMS. In addition, this paper provided a narrative account of the interdisciplinary research and multi-scale developments that recognize the value-addition of cutting-edge technologies in AI and computer vision.

Introduction

Roadway (pavement) infrastructure (RI) is considered to be the backbone of nation's economy [1]. Investment in roads has two intriguing dimensions: it improves the economy, while it is also an expensive process. Due to budgetary constraints, road maintenance activities are often postponed / delayed. Also, the delayed maintenance activities coupled with drastic increase in the traffic loading along with indeterminable climatic and environmental conditions not only lead to major structural and functional deterioration of pavement systems well before the end of their design lives, but also exponentially increase the cost of maintenance operations. Furthermore, if maintenance strategies are not accorded within the design service life, it would simply result in the failure of a pavement structure necessitating major rehabilitation / reconstruction, which is uneconomical [[2], [3], [4], [5]].

In 2019, American Society of Civil Engineers (ASCE) released the 2017 infrastructure report card for the United States of America (USA). The RI in California was ranked with a D+ grade representing the critical condition of pavement system. Similarly, the USA was ranked with a D grade construing the need to strengthen the RI [6]. The report stated that 68% of the major roads and highways in the USA are in poor condition and need to be maintained immediately. Also, it was estimated that US$61 billion was spent by the commuters annually due to increased vehicle operating costs, traffic delays, and incidents. In addition, an amount of US$130 billion was needed to bring back the pavements' condition from poor to good. This state was due, in large part, to delayed maintenance and deficient investment in upgrading the RI systems [1,6].

In order to reduce the severity of deterioration of roadway systems, and make strategic decisions regarding maintenance activities, highway agencies have historically focused on developing classic pavement asset management system (PAMS), which is basically a set of tools and systematic processes that help in planning, analyzing, and maintaining pavements at the right time with the right quality of materials in conjunction with budget availability. PAMS includes several modular systems such as data collection units, performance models, set of alternatives, decision making tools, implementation kits, and monitoring and feedback forms [7]. Albeit, the decisions taken through PAMS are widely affected by the method of data collection, analytical procedures, budgetary constraints, and influence of stakeholders.

Data collection is one of the most crucial systems, which affects the system efficiency. The dynamism of a database is consequential of rendering other dependent systems as highly functional allowing for betterment in data storage and retrieval, furthering the need for frequent updating of the existing database to reflect the changes in the pavement structure. Over the past few years, agencies have employed dedicated pavement data collection vehicles (DCV) to automate the data collection process. Though automated condition surveys provide high quality data, the process is costly [[8], [9], [10], [11], [12]]. DCVs are equipped with high-speed digital cameras, laser systems, and accelerometers that display significant sophistication in data collection procedures. Also, the unmanned aerial vehicles (UAV) equipped with camera modules and smart phone-based data collection methods have reduced the data collection costs [13,14]. Besides, these methods are capable of providing abstract visible pavement surface anomalies such as cracking, potholes, patching, raveling, and roughness.

Recent developments in the field of image processing (IP) and machine learning (ML) have provided cost-effective data analysis solutions [1,15]. IP & ML algorithms have been used to detect and quantify the distresses, along with assisting in computing condition indices and predicting future performance. Although these methods have provided reliable results, the computational costs are fairly high. Deep learning (DL), one of the subsets of Artificial Intelligence (AI) has significantly higher statistical assistance than other techniques, which also lowers computational costs. Therefore, the application of DL algorithms in PAMS data analysis has become one of the thrust areas to improve the analytical performance of PAMS. Recent studies in detecting and quantifying cracks from two-dimensional (2D) pavement images using DL algorithms have showed promising results [1,13,14]. However, applications of the cost-effective data collection and analysis methods have not been extensively tested for network-level PAMS activities, globally.

Each of the existing collection of literature has mainly focused on one specific aspect out of the several modules that encompasses data collection procedures, crack detection methods, and performance prediction models. However, a comprehensive document that addresses as many modules of PAMS as practical is necessary so new frameworks covering all aspects may be incorporated in the PAMS toolkit in future, which will help recommend solutions to improve the efficacy of the current roadway network.

Thus, the objective of this review paper was to collect and report several current state-of-the-art developments in PAMS and the associated embedded processes, majorly focused on data collection procedures, analytical techniques, decision making tools, and processing methods. In addition, this paper provides a narrative account of the interdisciplinary research and multi-scale developments that recognize the value-addition of cutting-edge technologies in computer vision and AI, eventually leading to the all-round development of futuristic PAMS. Further, this review document is envisioned to help link the various levels of PAMS and devise a globally acceptable system that fosters scientific and analytical formulations in a concurrent manner. The scope of this state-of-the-art paper covers the following: (i) description of the state-of-the-art developments in data collection procedures (Section II), (ii) detailed review of analytical techniques along with performance prediction and management decisions (Sections III), (iii) highlights of PAMS at global level (Section IV), (iv) challenges of incorporating soft computing methods in PAMS (Section V), and (v) elucidation of critical research gaps in the aforementioned processes that are essential parts of PAMS and recommendation of framework that encompasses creation of future roadmap in the areas of PAMS in order to improve the efficiency of the classic toolkit used to manage roadway assets (Section VI).