Research on Affordable Road Pavement Inspection for Municipalities Utilizing Smartphone LiDAR and LLM

Abstract

Pavement damage can lead to vehicle damage and injuries to road users; therefore, road management should shift from reactive maintenance to preventive maintenance. However, considering the workload of road administrators, implementing preventive maintenance is difficult, and pavement damage is often detected only after reports from citizens.

On the other hand, smartphone LiDAR has been applied in small-scale construction projects, suggesting its potential applicability to pavement inspection. In addition, the use of large language models (LLMs) for pavement inspection is expected to improve operational efficiency and provide technical support for municipal staff responsible for road management.

Therefore, this study aims to develop a low-cost pavement inspection method by combining smartphone LiDAR and LLM. The accuracy of crack ratio estimation was evaluated to assess the feasibility of the proposed approach. The results indicate that the estimation accuracy achieved at least the level corresponding to Condition Rating II+. Furthermore, by excluding data containing noise such as human shadows, accuracy corresponding to Condition Rating III was also achieved. Overall, the proposed method showed generally satisfactory accuracy.