Cloning Safe Driving Behavior for Self-Driving Cars using Convolutional Neural Networks

摘要

Background: In this paper, a Convolutional Neural Network (CNN) to learn safe drivingbehavior and smooth steering manoeuvring, is proposed as an empowerment of autonomous drivingtechnologies. The training data is collected from a front-facing camera and the steering commandsissued by an experienced driver driving in traffic as well as urban roads.Methods: This data is then used to train the proposed CNN to facilitate what it is called “BehavioralCloning”. The proposed Behavior Cloning CNN is named as “BCNet”, and its deep seventeen-layerarchitecture has been selected after extensive trials. The BCNet got trained using Adam’s optimizationalgorithm as a variant of the Stochastic Gradient Descent (SGD) technique.Results: The paper goes through the development and training process in details and shows the imageprocessing pipeline harnessed in the development.Conclusion: The proposed approach proved successful in cloning the driving behavior embedded inthe training data set after extensive simulations.