Uncertainty Aware Routing in Stochastic Transportation Networks with Correlated Link Travel-Times

该项目的目标是开发数学模型和算法，以实现大规模的路线规划工具，考虑到道路网络上行驶时间的不确定性。 该项目应对两大挑战。第一个是开发将大量历史车辆轨迹数据（例如，车辆在什么时间在哪里）转换为紧凑数据表示的方法。即将开发的方法将考虑到实时信息（如天气状况）、交通拥堵的传播（事故使附近路段减速）和用户偏好（一些司机比其他司机不那么激进）。第二阶段涉及开发路线规划算法，该算法将有效地集成这些丰富的数据模型，并为用户提供先进的路线规划帮助。该项目预计将通过提供本科生研究机会，行业实习以及研究生课程和项目的研究项目产生强大的教育影响。推广活动将针对妇女和代表性不足的少数民族（包括针对高中生的计划），让他们接触到尖端的交通研究。实现大规模的不确定性意识的路线规划应用程序需要在概率旅行时间预测领域的进步，使用大规模基于GPS的车辆轨迹数据和随机路由算法的网络与相关的链接旅行时间。该项目旨在解决交通网络背景下大规模数据建模和路由算法的交叉问题。从技术的角度来看，这需要开发新的模型，工作流程和算法，跨越交通建模和预测，机器学习和算法工程领域。该项目将重点关注在现实世界的应用背景下开发计算易处理的解决方案，目标是通过行业合作在实践中部署这些工具。预计该项目的成功完成将导致进一步的研究，如何可用的路线规划工具可以反馈到系统和影响网络的性能。这个奖项反映了NSF的法定使命，并已被认为是值得通过评估使用基金会的智力价值和更广泛的影响审查标准的支持。

The goal of this project is to develop mathematical models and algorithms for enabling large-scale route planning tools that account for the uncertainty of travel-times on road networks. The project addresses two major challenges. The first is to develop methods for converting very large amounts of historical vehicle trajectory data (e.g. where vehicles are at what times) into compact data representations. The approaches that are to be developed will account for real-time information (such as weather status), propagation in traffic congestion (accidents slow down nearby road segments) and user preferences (some drivers are less aggressive than others). The second stage involves developing route planning algorithms that will efficiently integrate these rich data models, and provide users with advanced route planning assistance. The project is expected to have strong educational impacts by providing undergraduate research opportunities, internships with industry, and research projects for graduate courses and projects. The outreach activities will target women and underrepresented minorities (including programs targeted at high school students) to expose them to cutting edge transportation research.Enabling large-scale uncertainty aware route planning applications requires advances in the areas of probabilistic travel-time predictions using large-scale GPS-based vehicle trajectory data and stochastic routing algorithms for networks with correlated link travel-times. This project aims to solve a problem at the intersection of large-scale data modeling and routing algorithms in the context of transportation networks. From a technical perspective, this requires developing new models, work-flows and algorithms that span the areas of traffic modeling and prediction, machine learning, and algorithm engineering. The project will have a strong focus on developing computational tractable solutions, in the context of real-world applications, with the goal of enabling the deployment of these tools in practice through industry collaborations. It is expected that successful completion of the project will lead to further research on how the availability of such route planning tools can feed back into the system and influence network performance.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.