Guiding Self -Driving Car Using Behavioral Cloning

Convoluted Neural Networks (CNNs) are widely adopted for pattern recognition tasks in image analysis. These algorithm captures the features in images using convolution operations with relatively fewer parameters in comparison to total number of operation. Additionally, they can implemented on massively parallel processing units like GPU’s to accelerate learning and inference. This project, implemented with CNN’s, illustrates the concept of transferring human driving behavior to machines or Self-Driving cars. The Objective of the project is to train a neural network to guide a Self-Driving car around a track in a simulator. During the training Phase, the model learns the steering angles for different position of car in relation to track. The acquired knowledge will then be used to predict the steering angles during the drive phase. The implementation is based on the Nvidia paper and uses Keras / Tensorflow deep learning library. <a href="https://medium.com/@madhusudhan.d/guiding-self-driving-car-using-behavioral-cloning-9c24541a425d">Click Here</a>