Work

RoadSentinel AI: A Predictive Framework for Road Damage Monitoring, Safety Risk assessment, and Maintenance Prioritization

Year: 2026
Type: article
Abstract: Road deterioration, including cracks, potholes, and pavement deformation, creates significant challenges for transportation safety and infrastructure management. Conventional road inspections often depend on manual surveys and periodic assessments, which can require considerable time, labour, and financial resources. This paper reviews existing research on deep-learning-based road-damage detection, pavement-crack assessment, lightweight object-detection models, machine-learning-based pavement-condition prediction, and road-damage datasets. Based on the identified research gaps, this paper proposes RoadSentinel AI, an integrated predictive road-monitoring and maintenance decision-support framework. The proposed framework combines image- and video-based damage observations with historical road-condition records and, where available, traffic, weather, road characteristics, location, and maintenance information. The intended capabilities include: Road damage detection and classification. Damage severity assessment. Historical road-condition monitoring. Road health scoring. Deterioration-risk estimation. Probable contributing-factor analysis. Road safety-risk assessment. Maintenance prioritization and recommendations. The proposed framework aims to connect automated visual inspection with historical condition analysis, risk assessment, and road maintenance planning. This research presents a conceptual framework rather than an experimentally validated implementation. Future work will focus on system development, collection of historical road-condition data, validation of prediction and scoring methods, and evaluation in real-world road environments. Institution: MIT ADT University, Pune, India.
Language: English
Open Access status: green
OpenAlex ID: W7236830876
FWCI:
Cites:
Cited by:
Related to: 10