Advanced traffic flow theory and control for heterogeneous intelligent traffic networks

该奖学金将开发新一代基于实时模型的控制框架，使工程师能够通过主动交通管理（ATM）程序管理和控制异构智能交通系统的实时操作。一般来说，ATM程序，也被称为管理车道或智能车道，是一种改善交通流量和减少高速公路拥堵的计划。它利用自动系统和人工干预来管理交通流量，确保道路使用者的安全。信息和通信技术（ICT）已经改变了商业、社会和政府的许多方面，从医疗保健到教育和经济。信息和通信技术目前正处于改变运输系统的早期阶段，它将传感器（遥感和定位）、控制单元（交通信号、信息标志）和自动化技术与微芯片结合起来，使它们能够通过无线技术相互通信。在许多发达国家，特别是日本和韩国，在ATM计划中部署ICT已经导致交通网络性能的显著改善，例如减少拥堵，提高交通安全，提高环境质量（例如减少CO2）以及为道路使用者提供更可靠的服务。预计在未来5至10年内，ICT将在全球范围内取得显著进展，智能装备车辆将在交通流量中占据重要份额，其中驾驶任务通过自主车辆到车辆（V2V）和车辆到基础设施（V2I）通信从驾驶员转移到车辆。在V2V通信中，领先的装备车辆将向更上游的车辆发布其当前速度、驾驶操纵（例如，加速或减速）等的信息，而在V2I通信中，装备车辆将与路边智能设备交换信息并从这些设备接收命令以进行随后的驾驶活动。交通流中相当大比例的智能车辆将形成包含非装备（或手动）和装备车辆混合组成的智能交通网络。这样的交通流系统被定义为异构智能交通系统。本研究旨在寻求一种改进的ATM方案，以更有效地监控智能交通网络。原则上，异构智能交通网络的交通控制问题是高度复杂的，其特征在于非装备车辆与各种装备车辆之间的相互作用以及装备车辆与路边智能设备之间的相互作用，以及不同类型车辆的不同控制策略之间的相互作用。拟议的研究将解决这些复杂的问题，并引入一个新的基于实时模型的智能交通控制框架，使用从多个来源（环路探测器，遥感，移动的电话，浮动汽车等）收集的真实数据。新模型将在短期内预测交通拥堵模式（即自由流，拥堵或停止和去堵塞之间的过渡），并调查这种拥堵的真正原因，发生在异构智能交通网络。根据实时模型预测的交通状态，将为不同类型的车辆（配备和未配备）建立一系列即时控制措施，以减少拥堵，旅行时间和空气污染。

This fellowship will develop a new generation of a real-time model based control framework required for engineers to manage and control the real-time operations of a heterogeneous intelligent traffic system through Active Traffic Management (ATM) programs. In general, an ATM program, also known as managed lanes or smart lanes, is a scheme for improving traffic flow and reducing congestion on motorways. It makes use of automatic systems and human intervention to manage traffic flow and ensure the safety of road users.Information and communication technologies (ICT) have transformed many aspects of business, society and government, from healthcare to education and the economy. ICT are now in the early stages of transforming transportation systems by integrating sensors (remote sensing and positioning), control units (traffic signals, message signs) and automatic technologies with microchips to enable them to communicate with each other through wireless technologies. In many developed countries, particularly Japan and South Korea, the deployment of ICT in ATM programs has led to significant improvement of traffic network performance such as reduced congestion, increased traffic safety, enhanced environmental quality (e.g. reduced CO2) and a more reliable service to the road user. It is expected that in the coming 5 to 10 years ICT will considerably progress worldwide so that intelligent equipped vehicles, in which the driving tasks are shifted from the driver to the vehicle through autonomous vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication, will make up a significant share of the traffic flow. In V2V communication, the leading equipped vehicle will issue information of its current speed, driving manoeuvre (e.g. acceleration or deceleration), etc. to further upstream vehicles while in V2I communication, the equipped vehicle will exchange information with roadside intelligent devices and receive commands from such devices for consequent driving activities. A considerable proportion of intelligent vehicles in traffic flow will create intelligent traffic networks containing a mixed composition of non-equipped (or manual) and equipped vehicles. Such traffic flow system is defined as a heterogeneous intelligent traffic system. This proposal will seek solutions for an improved ATM program to monitor and control more efficiently intelligent traffic networks.In principle, the traffic control problem for heterogeneous intelligent traffic networks is highly complex, which is characterized by the interactions between non-equipped vehicles and various types of equipped vehicles and by the interaction between equipped vehicles and the roadside intelligent devices, as well as by the interplay between different control strategies for different types of vehicles. The proposed research will tackle such complex issues and bring in a new real-time model-based intelligent traffic control framework using real-life data collected from multiple sources (loop detectors, remote sensing, mobile phones, floating cars, etc. ). The new model will predict in the short term the traffic congestion patterns (i.e. the transitions between free-flow, congestion or stop-and-go jams) and investigate the true causes of such congestion which occurs in a heterogeneous intelligent traffic network. Based on the traffic states predicted from the real-time model, a sequence of immediate control actions will be established for different types of vehicles (equipped and non-equipped) in order to reduce congestion, travel time and air pollution.