Intersection Traffic Efficiency Enhancement using Deep Reinforcement Learning and V2X Communications

초록

Despite of significant research on intersection traffic management and traffic efficiency, it is still a major issue due to increasing number of vehicles and new challenges created by autonomous and cooperative driving applications. At the same time, progresses in artificial intelligence (AI) and vehicle to everything (V2X) technologies are leading new opportunities to achieve optimal traffic efficiency. This paper addresses the intersection congestion issue using V2X communication and Reinforcement Learning (RL) based traffic control which is trained and evaluated through Simulation of Urban MObility (SUMO) simulator. The experimental results show waiting time at the intersection is reduced as the training progresses.