Simulation and Deep CNN based architecture for validation of Intelligent automotive functions

Intelligent automotive software applications such as advanced driver assistant systems and self-driving cars possess tremendous potential to revitalize the transportation sector. Software-enabled vehicles tend to be a game-changer on the roads, transforming the entire structure of driving which we see nowadays. The massive use of sensors and actuators along with the availability of large amounts of data are fostering artificial intelligence applications such as machine learning and deep learning algorithms to be extensively adopted in the transportation domain. Deep convolutional neural networks (CNN) can play a substantial role in driving vehicles autonomously without human intervention. Autonomous driving relies on the meticulous functioning of advanced driver assistance systems. This paper, however, describes an approach to utilize existing deep neural network principles to construct a mechanism for autonomous driving functions. A deep neural network algorithm has been proposed for an automated driving scenario. The designed algorithm is then evaluated on a simulator which produces highly effective results. Unity framework is used as a simulator for manually providing input data.