Visual Perception Deep Drive Model for Self-Driving Car

Abstract

Self driving cars are the need of future technology, there are many companies that are trying to perfect this particular project but there are still some deficiencies there. Most of the companies are using Expensive sensors like RADAR and LiDAR to get the idea of environment, which are very hard to use and need a lot of processing power. Our project focuses on using only visual aid to drive a car, particularly following the lane of the road. We trained a model using Convolutional Neural Network (CNN), in a simulated environment and tested the model in the same environment.