The state of the road surface has a big impact on passenger comfort, vehicle performance, and transportation safety. Potholes, fissures, and uneven surfaces are examples of poor road conditions that can cause collisions, car damage, and higher maintenance expenses. Conventional road monitoring techniques depend on specialised equipment or manual inspection, which are frequently costly and ineffective for extensive road networks. Road conditions can now be monitored using sensor data gathered from mobile devices thanks to the quick development of smartphones with integrated sensors like accelerometers and GPS modules. In this study, a machine learning-based method for identifying road surface conditions using sensor data from smartphones is presented. To categorise road quality, the suggested method makes use of features including vehicle speed, GPS position, accelerometer readings along three dimensions, and road type information. Thousands of records depicting various road conditions classified as Excellent, Good, Satisfactory, and Bad make up the dataset used in this study. To create categorisation models, two machine learning techniques are used: Support Vector Machine (SVM) and Extreme Gradient Boosting (XGBoost). According to experimental findings, the XGBoost model outperforms the SVM model in terms of accuracy, which makes it more useful for classifying road conditions. The suggested method offers a scalable and affordable way to monitor roads in real time with widely accessible smartphones. By enabling autonomous road quality detection and facilitating prompt road maintenance decisions, this method contributes to intelligent transportation systems.
Ekambaram et al. (Thu,) studied this question.