NSF/USDOT Partnership for Exploratory Research - ICSST: Towards a Systems Integration Urban Network Performance Measure - Scalability and Data Issues of the Two-Fluid Model

机构：内布拉斯加大学-林肯国家科学基金会/美国交通部：走向系统集成城市网络性能测量——双流体模型的可扩展性和数据问题为了推进地面交通系统信息技术的集成，本研究将试验利用从街道走廊的摄像头阵列收集的基于实时视频的交通流量数据，用于交通控制和城市交通系统建模。该研究旨在使交通控制系统能够与交通环境进行交互，基于感官输入获取和理解交通状况，并对交通控制任务的执行情况做出智能决策。拟议的研究将集中于以下目标。(1)视频切换和目标车辆跟踪；(2)交通流信息提取；(3)车辆驾驶模式建模和识别（例如，识别表现出危险行为或潜在风险的个别车辆）。为了实现这些目标，本项目的研究活动将分为三个任务阶段：(1)图像采集和预处理阶段，(2)车辆跟踪和移交处理阶段，(3)交通视频理解（车辆驾驶模式识别）阶段。

Elizabeth G. Jones Institution: University of Nebraska - LincolnNSF/USDOT: Towards a Systems Integration Urban Network Performance Measure - Scalability and Data Issues of the Two-Fluid ModelTo advance the integration of information technologies in surface transportation systems, this research will experiment with the utilization of real-time video-based traffic flow data collected from camera arrays along a street corridor for traffic control and city transportation system modeling. The research is expected to enable the transportation control system to interact with the traffic environment, to acquire and understand the traffic situations based on sensory inputs, and to make intelligent decisions about the circumstances where the traffic control tasks are to be performed. The proposed research will focus on the following objectives. (1) Video handoff and target vehicle tracking, (2) Traffic flow information extraction, and (3) Vehicle driving pattern modeling and identification (e.g., identifying individual vehicles that demonstrate dangerous behavior or potentially at risk). To achieve these objectives, the research activities of this project will be divided into three task phases: (1) the image acquisition and pre-processing phase, (2) vehicle tracking and hand-off processing phase, and (3) the traffic video understanding (vehicle driving pattern recognition) phase.