A Review on Contribution and Influence of Artificial Intelligence in Road Safety and Optimal Routing

Abstract

Pothole detection is crucial for road safety and maintenance, driving research towards automated and efficient detection systems. Traditional methods present limitations: public reporting, while cost-effective, relies on citizen participation and lacks real-time information; vibration-based methods, using accelerometers to detect vehicle vibrations, require driving over potholes. Image/video processing techniques offer a proactive approach by analysing visual data to identify potholes. These methods often leverage computer vision algorithms, 3D scene reconstruction, and machine learning strategies for enhanced accuracy. Researchers are exploring deep learning models like Convolutional Neural Networks (CNNs) and YOLOv2 to im- prove real-time pothole detection accuracy and efficiency. These advancements, including stereo vision-based systems with high detection rates and pixel-level accuracy, contribute to timely pothole detection and repair, ultimately improving road safety.