Self-moving multi-sensor AI-based robotic technology for road crack inspection
Résumé
Human inspectors conducting road inspections face both heavy physical demands and subjective judgment, which can affect the accuracy of road surface evaluations. This study introduces Mobi-A4Net, an affordable robotic-AI system for automatic road crack detection and assessment, addressing limitations of costly automated systems and lack of integration with existing technologies. The system focuses on an unmanned ground vehicle (UGV) platform for detailed inspection work. The Mobile Adaptive Attention Aggregation Network forms the core of the system, implementing a compact deep-learning model with multi-scale attention mechanisms to identify thin, low-contrast cracks within complex surface patterns. Advanced image-processing techniques, including Medial Axis Transform (MAT) skeletonization, enable real-time measurement of crack length, width, and orientation. Experimental results show that Mobi-A4Net achieves 99.7% detection accuracy, a recall rate of 98.8%, and a mean Intersection over Union (mIoU) of 95.4%. With 1.85 million parameters and an inference speed of 9.6 milliseconds per image, the system is suitable for real-time operation on embedded UGV platforms.
Citer ce document
Exporter : BibTeX · RIS (Zotero, Mendeley, EndNote)
Accès au document
Texte intégral archivé sur SPHAERO, en accès ouvert : lisible gratuitement en ligne après connexion. Le document reste aussi disponible à sa source. Se connecter
Voir l'article sur le site de la revueLicence et provenance
Licence : CC BY
Notice moissonnée depuis OpenAlex le 01/10/2026. Le document reste hébergé par sa source.
Voir le document à la source →