Applying Deep Learning to Autonomous Vehicles: A Survey

Abstract

Autonomous vehicles will revolutionize the future automotive industry by reducing or eliminating human errors and increasing vehicle safety. The design of autonomous vehicles will bring together many modern technologies such as control, computer vision, path planning, sensor fusion and fault diagnosis. Although traditional algorithms can be used to implement these technologies, in the last decade, deep learning has been proposed to challenge traditional algorithms and excels in many applications and research areas. Deep learning has shown its significant use in the design and operation of autonomous vehicles. In this paper, we will review the application of deep learning in five major research areas of autonomous vehicles i.e. path planning, control, computer vision, sensor fusion and fault diagnosis. Recent advances of the deep learning in autonomous vehicles will be reviewed and discussed.