详情

Establishing a Simulation Package and Testbed for Traffic Congestion Reduction Using Deep Reinforcement Learning

项目名称：	Establishing a Simulation Package and Testbed for Traffic Congestion Reduction Using Deep Reinforcement Learning
摘要：	Traffic congestion not just causes travel delays but also increases fuel consumption and emissions production [1]. One of the major reasons for congestion in urban areas is traffic accidents. Currently, traffic cameras and video surveillance are some of the ways used to monitor the traffic [2]. However, these methods are capital demanding and do not provide real-time trip information to the travelers. New technologies, such as vehicle to infrastructure (V2I) and vehicle to vehicle (V2V) communication, may be able to greatly reduce congestion. This type of communication allows real-time detection of congestion, which can result in immediate distribution of traffic affected by the congestion and therefore result in a more efficient transportation network. Advances in wireless communication technology, for instance, advanced 5G communication networks will enable this interconnection and will allow users to make better decisions regarding the use of the transportation system. In the foreseen transportation infrastructure, vehicles will communicate with other vehicles, traffic control units, and traffic management centers, to make more efficient trip decisions. All these technologies have paved a solid foundation for autonomous driving, which has been identified as a national priority for future technologies with an expected $488 billion in annual savings from reducing traffic accidents and another $158 billion in savings due to reduced fuel costs [3]. This project aims to develop a simulation package for autonomous driving and route redirection in a designated region using reinforcement learning (RL) algorithms. The developed RL algorithms will determine the motion and routes of vehicles considering the shortest traveling path, shortest traveling time, and traffic conditions to reduce traffic congestion. The research team will further verify the algorithms using a hardware-in-the-loop testbed including scale-down tracks, car-like rovers, and traffic signaling systems. As more and more traffic-related data has been collected and deposited in the past decades, the application of data-driven artificial intelligence to solve traffic congestion is one of the most promising approaches for intelligent transportation systems. This project will advance two national priority areas in research: artificial intelligence and autonomous driving. All outcomes of the project will be shared with DOT and 3rd party stakeholders to benefit the community at large. The proposed project will also seamlessly integrate the booming research on RL with educational activities by training undergraduates with senior design and research experiences for the undergraduate programs at UTSA, training graduate students with thesis and dissertation projects, and high school students with a summer training program to prepare the future workforce for intelligent transportation systems.
状态：	Active
资金：	106000
资助组织：	Office of the Assistant Secretary for Research and Technology
管理组织：	Transportation Consortium of South-Central States (Tran-SET)
项目负责人：	Mousa, Momen R
执行机构：	University of Texas at San Antonio
主要研究人员：	Jin, Yufang
开始时间：	20220401
主题领域：	Data and Information Technology；Highways；Operations and Traffic Management；Planning and Forecasting；Safety and Human Factors；Vehicles and Equipment