Advanced Interaction Model for autonomous transportation system using Deep Learning Methods

Abstract

Deep learning (DL) plays a major role for advancing transportation systems. Recently, the researchers have witnessed the advent and prospect of deep learning which has become a hot topic in ITS (Intelligent Transportation Systems). As a result, traditional learning models in many applications made a way for deep learning for its new learning techniques so that the landscape of ITS can be reshaped. Autonomous vehicles promise to improve road safety to protect the traffic congestion meanwhile, an increase fuel usage. This paper introduces how to plan advanced transportation system with autonomous vehicles in traffic. We model the interaction between the autonomous vehicle with the surrounded roads with Deep Inverse Reinforcement Learning (DIRL).  We validate the proposed application with maximum entropy principle (MEP) to learn the effectives of proposed model. Simulated results prove that expected driving behaviors of an autonomous vehicle with DIRL gets the promising and competitive results compared with other works.