CAREER: Modeling and Estimation Methods for Complex Traffic

这个学院早期职业发展(Career)项目奖的目标是调查复杂交通的动态。复杂交通的特征是不同类型的车辆(如自行车和汽车)，它们的大小和性能特征各不相同，但共享相同的基础设施，并且通常由人类控制。在特殊活动导致的极端拥堵期间，这些特征在美国越来越常见，在世界各地的新兴经济体中也很普遍。这项研究假设，数学模型的进步、大量交通数据的提供和验证是充分理解复杂交通的关键因素。这项研究的重点是(I)异质交通的数学模型的建立，(Ii)人工交通的建模和分析，以及(Iii)快速准确的估计算法的发展，以将数据整合到城市规模的模型中。通过新开发的交通感知技术获得验证模型和估计算法的数据，如果成功，这项工作将为下一代交通监控和管理系统的开发提供支持。最终，这将有助于降低美国在特殊活动期间每年数十亿美元的拥堵成本。开展教育和外联活动，使学生具备设计下一代计算机增强型基础设施所需的计算能力。这是通过针对本科生和研究生的新的教育举措来实现的，这些举措强调将编程和计算技能应用于土木工程问题。交通界的可重复计算研究计划将有助于最大限度地发挥研究的潜在影响，并增加从业者采用该研究的可能性。更广泛的社区参与计算机在交通运输中的应用是通过外展和开放的课件活动实现的。

The objective of this Faculty Early Career Development (CAREER) program award is to investigate the dynamics of complex traffic. Complex traffic is characterized by heterogeneous vehicle types (e.g. bikes and cars) that vary in size and performance characteristics but share the same infrastructure, and is often controlled by humans. These features are increasingly common in the US during extreme congestion generated by special events, and are pervasive in emerging economies worldwide. This research postulates that advances in mathematical models, informed by and validated with large volumes of traffic data, are key elements to unlock the full understanding of complex traffic. This research focuses on (i) the development of mathematical models of heterogeneous traffic, (ii) modeling and analysis of human-directed traffic and (iii) the development of fast and accurate estimation algorithms to integrate data into city-scale models. Data to validate the models and estimation algorithms are obtained through a newly developed traffic sensing technology.If successful, this work will support the development of next generation traffic monitoring and management systems. Ultimately, this will help reduce the multibillion-dollar annual cost of congestion during special events in the US. Educational and outreach activities are executed to prepare students with the computing competencies needed to engineer the next generation of computer enhanced infrastructure. This is achieved through new educational initiatives for undergraduate and graduate students that emphasize programming and computational skills applied to problems in civil engineering. Reproducible computational research initiatives within the transportation community will help maximize the potential impact of the research and increase likelihood of adoption by practitioners. Engagement of the broader community on applications of computing in transportation is achieved through outreach and open courseware activities.