|
Original research SMART POTHOLE DETECTION AND ADAPTIVE VEHICLE RESPONSE A CNN AND IOT DRIVEN APPROACH FOR ENHANCED ROAD SAFETY AND MAINTENANCEPages 13-24
Abstract
Pothole detection and mitigation systems are integral to enhancing road safety and optimizing vehicle performance. By leveraging Convolutional Neural Networks (CNN) and Internet of Things (IoT) technologies, these systems enable real-time hazard identification, such as potholes, and facilitate dynamic vehicle adjustments to mitigate potential damage. This study analyzed over 1,000 pothole images, with the CNN model achieving an accuracy rate of 96% to 99%. Edge detection techniques, including Sobel filters, were utilized to assess key pothole attributes such as diameter, depth, and edge sharpness. For potholes with diameters exceeding 50 cm and edge sharpness above 85%, the vehicle's suspension damping was automatically increased by 40%, minimizing the impact on the vehicle's chassis. Additionally, the system dynamically reduced vehicle speed by 10–20 km/h for severe potholes, based on real-time analysis by the Electronic Control Unit (ECU). The ECU also communicated with the Anti-lock Braking System (ABS) to apply braking force when sharp-edged potholes were detected. In scenarios where rear vehicles maintained a safe distance of 50 meters, the braking system was activated, reducing the risk of tire damage and collisions. Through IoT integration, real-time data was stored in the cloud, enabling predictive maintenance and improving repair planning efficiency by 30%. This approach not only enhances passenger safety but also reduces vehicle wear and tear, while improving road infrastructure management efficiency. The combination of CNN and IoT-based solutions marks a significant advancement in automotive safety systems.
Keywords: Pothole Detection; Convolutional
Neural Networks (CNN); Internet of Things (IoT); Edge Detection; Automotive Safety Systems.
|
Krishnaraj J., 
PDF 