CRII: CPS: Emerging Markets and Myopic Decision-Making in Multi-Modal Transportation Systems: Modeling and Validation

该项目旨在创建高保真模型，并通过真实世界的数据进行验证，以了解混合模式的出行决策和新兴的移动市场。越来越多的旅行者通过应用程序做出决策，这些应用程序根据用户定义的偏好优化（混合模式）路线。本地优化的解决方案往往会导致效率低下，而旅行者的风险敏感性（源于内源性和外源性不确定性）又会加剧这种情况。传统的理性的、效用最大化的模型往往无法捕捉到这些影响，特别是在短期决策中，因为这些决策几乎没有时间进行思考，而参考点在选择中起着主要作用。该项目旨在（a）学习考虑风险敏感性的旅行者决策模型，（B）开发市场结构模型（例如，乘车共享平台），其捕获出行者对移动模式的评价。该研究利用前景理论来捕捉风险敏感性和有限理性来捕捉信息不对称和近视，从而扩展了旅游决策的理性模型。这些概念将被集成到反向强化学习框架中，我们寻求在线算法来支持控制/激励设计。此外，我们将开发技术方法来建模移动市场结构，包括捕捉风险敏感性的新排队博弈理论模型。拟议的工作将受益于通过市政和行业合作伙伴提供的多式联运数据，以便在一个完善的测试平台内进行测试和验证。该项目将突出市政当局可以调整其管理战略的领域，以补充共享经济，以公平的方式解决服务差距。通过与西雅图交通部和行业合作伙伴Swiftly和IDAX的合作，有可能将结果转化为实践。研究生和本科生将积极参与该项目，并将有充分的机会通过我们的市政合作伙伴与社区接触，并接触到未来智慧城市的工作。

This project aims to create high-fidelity models, validated with real-world data, of mixed-mode travel decisions and emerging mobility markets. A growing subset of travelers make decisions informed by apps that optimize (mixed-mode) routes based on user-defined preferences. Locally optimized solutions tend to cause inefficiencies that are exacerbated by risk-sensitivity (arising from endogenous and exogenous uncertainties) in travelers. Traditional rational, utility maximization models tend not to capture these effects, particularly in short-horizon decisions that leave little time for cogitation and points of reference play a primary role in choice. The project aims to (a) learn models of traveler decision-making that account for risk-sensitivity and (b) develop models of market structures (e.g., ride-sharing platforms) that capture traveler valuations of mobility modes. The proposed research extends rational models of travel decisions by leveraging prospect theory for capturing risk-sensitivity and bounded rationality for capturing information asymmetries and myopia. These concepts will be integrated into an inverse reinforcement learning framework for which we seek online algorithms in support of control/incentive design. Moreover, we will develop technical approaches to modeling mobility market structures that include new queuing game-theoretic models that capture risk-sensitivity. The proposed work will benefit from multi-modal transit data available via municipal and industry partners for testing and validation within a well-developed testbed.This project will highlight areas where municipalities can adjust their management strategies to supplement the sharing economy in addressing the service gap in an equitable way. Through collaborations with the Seattle Department of Transportation and industry partners Swiftly and IDAX, there is potential to translate results to practice. Graduate and undergraduate students will be heavily involved in the project and will have ample opportunity to engage with the community through our municipal partners as well as be exposed to the jobs of tomorrow's smart cities.