Pothole detection on Roads using CNN for Autonomous vehicles

Pothole is a large fissure while the vehicles are on the roadways. The potholes usually lead to underneath caverns and pits in the road to make it difficult to ride. If the vehicles are driven by humans, after seeing them, they move away from that path but when it comes to self-driving or autonomous vehicles, it has to be detected. Autonomous vehicles use a variety of technologies to drive themselves such as "autopilot." In Autopilot mode the driver can specify the initial point and designation point, as it automatically redirects the route for destination location. Additionally, autonomous vehicles help to reduce accidents. Pothole detection is a process that begins with data gathering from roadways. After gathering data from the real world, data preprocessing is done to clean and remove unwanted data. the proposed methods uses convolutional neural networks (CNN) algorithm with various architecture such as alexnet CNN model, Lenet architecture, Manet architecture are used. The comparison of all those models are obtained. To Identify the plain and pothole were implemented using internet of things with accuracy about 87.5% as result.