Decision System for a Self-Driven Car

Abstract

As the development in the sensor technology and as the accuracy of those sensors keep on upgrading, the perception of the surrounding environments and operational intricacy of the connected automated vehicles also enhances. In this paper, we propose an implementation of a decision-making system which is being developed for the control of autonomous vehicles. It focuses on the efficient usage of neural networks to realize the classification of traffic signs, detection of vehicles and the ultimate aim of a self-driven car that is to drive over any path without any human intervention. Unity based Udacity simulator can be modified to make variety of paths available to train and test the built autonomous car.