Secure Communication and Pothole Detection for UAV Platforms

Abstract

Everyone around the world relies on roads for transportation on a daily basis. Aging and heavy usage of roads results in deterioration of the surface. Besides potholes, we also see many cracks in roads which may expand and develop into potholes themselves. Potholes can result in vehicle damage and can also cause physical harm to people in their vehicles. Detected potholes, if integrated with information systems, can notify drivers in real time, so they can be aware. Potholes are a problem, especially in developing countries. In this paper, we discuss leveraging and integrating the best modern technologies in order to solve this problem. We have tried to make it a feasible, robust, flexible, and modular system to help solve this problem. Using unmanned aerial vehicles (UAVs) and machine learning/deep learning and networking together, we will be able to spot potholes and cracks on the roads much faster than existing methods. By using the system, the need for manual surveillance is reduced. Using a UAV, you can cover large areas. To detect potholes, a UAV will be used to take videos and take pictures along the roads. These images will be sent to a computer where ML algorithms will detect the pothole. In addition to the CNN, a transfer learning algorithm and YOLO real-time object detection can be used to identify potholes. Video is streamed securely over Wi-Fi by an RTSP server to a local workstation, so the user can view the video from the UAV and may store it for future use.