Speed Bump Detection on Roads using Artificial Vision

Abstract

In recent decades, self-driving has been a topic of wide interest for Artificial Intelligence and the Automotive Industry. The irregularities detection on road surfaces is a task with great challenges. In developing countries, it is very common to find un-marked speed bumps on road surfaces which reduce the security and stability of self-driving cars. The existing techniques have not completely solved the speed bump detection without a well-marked signaling. The main contribution of this work is the design of a methodology that use a pre-trained convolutional neural network and supervised automatic classification, by using the analysis of elevations on surfaces through stereo vision, for detect well-marked and no well-marked speed bumps to improve existing techniques.