A Global Perspective on Road Condition Assessment and Maintenance: Trends in Research and Technology Integration

• Bibliometric review of pavement condition assessment and maintenance research from 1990 to 2025. • Publications show steady growth with a marked increase after 2015 linked to digitalisation and sustainability agendas. • PCI, IRI, and cracking indices dominate assessment, while AI, IoT, and imaging methods reshape predictive analysis. • Research output is led by the United States, China, and Canada, with increasing contributions from emerging economies. • Gaps remain between academic advances and field implementation, highlighting the need for applied validation. This study conducts an extensive bibliometric review to evaluate the global evolution of pavement condition assessment and maintenance research over the last three decades. Using a dataset of 634 Scopus-indexed publications, the study applies co-authorship, citation analysis, keyword co-occurrence, bibliographic coupling, and text-based mapping to explore scholarly productivity, thematic clusters, and technological routes. Findings highlight a growing concentration around core evaluation indices such as PCI and IRI, with newer studies embracing machine learning, remote sensing, and decision-support systems. Research leadership remains concentrated in countries with advanced road infrastructure, particularly the United States, China, and Canada although contributions from emerging economies are steadily growing. Despite technological progress, a gap still exists between academic innovation and practical application. Only a small portion of the literature has influenced real-world practice, mainly because of limited resources, fragmented data, and a lack of interdisciplinary collaboration. Case studies from the Washington State Department of Transportation (DOT) and Maine DOT illustrate effective implementation of data driven and AI-enabled pavement systems. This underscores the potential of AI to automate condition assessments, enhance predictive modeling, and optimize rehabilitation strategies. However, the study also reveals key limitations, including limited model generalizability, challenges in reproducibility, and varying levels of stakeholder readiness. Future research should focus on developing open-access datasets, collaborative validation frameworks, and practical toolkits that can be readily adopted by road agencies. Moreover, integrating AI into research workflows offers valuable opportunities to connect theory with practice, creating smoother pathways from academic innovation to real-world implementation and ensuring that research delivers tangible, lasting impact.