ScholarWorks@숙명여자대학교

초록

As the development of technology advanced, the number of vehicles increased significantly. That is why traffic congestion is an inevitable problem. Traffic management is essential to alleviate traffic congestion and improve traffic flow. Traffic management is largely divided into signal control and trajectory control. Among them, we propose DRL(Deep Reinforcement Learning)-based trajectory control methods with different objectives to improve traffic flow. In different environment structures, we compare and analyze the performance of traffic flow in terms of travel time, travel distance and the vehicle density for optimal real-time trajectory control.

키워드