Optimizing Future Mobility Systems

这项研究计划的目的是研究未来移动系统的建模、设计和运行中出现的优化问题，这带来了几个主要与潜在问题的规模有关的挑战。在建模方面，挑战在于开发综合了天气条件、能源价格等多种因素以及潜在不确定性的模型。在方法论方面，挑战在于开发能够处理问题的规模和不确定性的大规模优化技术。 未来移动系统：联网汽车、混合动力汽车和电动汽车以及目前市场渗透率较小的自动驾驶汽车的新创新预计将在不久的将来成为主流。这些新技术将使货物和人员的运输变得更便宜、更可持续。此外，市场正在从传统的私家车拥有向移动即服务转型，其中私家车拥有与共享汽车服务共存。这样的系统需要解决几个具有挑战性的数据驱动的优化模型，以最终运行高效和可持续的移动系统。 数据驱动的模式：对于大多数人来说，出行模式每天都是一样的，比如从家到单位。通过查看来自不同来源的数据，数据驱动模型可以了解每个人的出行模式，从而创建一个系统，该系统可以看到人们在哪里，检查他们正朝哪个方向移动，然后预测他们将去哪里。此外，通过考虑交通和能源价格预测，混合动力和电动汽车的路线选择可以这样一种方式进行，以最大限度地减少旅行成本和对环境的影响。例如，在混合动力汽车的情况下，电动发动机和内燃机的使用可以通过这样的方式进行调度，以限制在人员高度集中的地点排放温室气体。 大规模分布式优化：电力使用的有效调度以减少对环境的影响，以及TRIPS的路线和调度，应该在全系统范围内进行。因此，将要开发的优化模型本质上是大规模的优化模型，在计算上具有挑战性。拟议的研究计划将研究可伸缩大规模优化的新技术，这将需要对优化问题的基本结构有更深的理解。将研究的优化方法包括分解、列生成和切割平面，以及分布式实施，该实施使建议的解决方案方法可扩展以用于云平台上的实际部署。

The aim of this research program is to investigate optimization problems arising in the modeling, design, and operation of future mobility systems, which entails several challenges mainly related to the scale of the underlying problems. On the modeling side, the challenge is in developing models that integrate a plurality of factors such as weather conditions and energy prices along with the underlying uncertainty. On the methodological side, the challenge is in developing large scale optimization techniques that can handle the scale and the uncertainty of the problems. Future Mobility Systems: New innovations in connected vehicles, hybrid and electric vehicles, and autonomous vehicles which currently have small penetration in the market are expected to become more mainstream in the near future. These new technologies will enable cheaper and more sustainable transportation of goods and people. Furthermore, the market is transforming from the traditional private vehicle ownership to mobility-as-a-service, where private vehicle ownership co-exists with shared vehicle services. Such systems entail the solution of several challenging data driven optimization models to ultimately operate an efficient and sustainable mobility system. Data Driven Models: For most people, travel patterns are the same every day, such as traveling from home to work. By looking at the data generated from different sources, data driven models can learn the travel patterns of each individual and thus create a system that sees where people are, examines in which direction they are moving, then predicts where they will go. Furthermore, by taking into account traffic and energy prices predictions, the routing of hybrid and electric vehicles can be performed in such a way to minimize the cost of travel and the impact on the environment. For instance, in the case of hybrid vehicles, the usage of the electric engine and the combustion engine can be scheduled in such a way to limit the release of green-house gas in locations with high concentration of people. Large Scale Distributed Optimization: The efficient scheduling of electric energy usage to reduce the impact on the environment, along with the routing and the scheduling of trips, should be done on a system-wide approach. The optimization models that will be developed are thus essentially large scale optimization models that are computationally challenging to solve. The proposed research program will investigate new techniques for scalable large scale optimization which would require a deeper understanding of the underlying structure of the optimization problems. The optimization approaches that will be investigated include decomposition, column generation, and cutting planes along with a distributed implementation that makes the proposed solution methods scalable for real life deployment on a cloud platform.