Comparative Analysis of CNN models for Self-Driving Cars in a Simulated Environment

Casualties due to human failure in a road accident increases with every passing month and with technology assisting humans in every possible field, it is essential to prioritize road safety with it. A self-driving car is an emerging subject for every automobile giant in the world. Autonomous cars not only help in reducing human errors while driving but also help in reducing driver fatigue to a greater extent. This paper uses a more budget-friendly yet effective option using a model based on Nvidia’s CNN Model to make the car to detect lane lines and to steer the car in a virtual simulator so that it remains bound to them. The paper compares two proposed models one is based on Nvidia’s existing CNN model (model l) and the other is proposed CNN model (model 2). Modell achieved an outstanding accuracy of 97.22% with the correct set of input parameters while the other proposed model had an accuracy of 88.35%. Both the models were also compared to similar models from an existing work in literature, and it was observed that in comparison the proposed model had better accuracies.