Making use of computer vision approach for implementation of self-driving car

Abstract

This article presents an approach of steering a self-driving car on the road constructed from traffic cones using Computer Vision. Our algorithm gets an image in real-time from an onboard-camera and then, depending on the results of the analysis, it calculates a steering angle. After training, our model could forecast how and when to steer the self-driving car in real time. As a result, we reached the accuracy of steering about 95%.