Integrated Multimodal Travel Behavior Analysis under Mobility as a Service

移动即服务(MAAS)是城市交通可持续发展的一种重要手段，近年来在全球交通领域备受青睐。通过集成多个提供商的服务(例如出租车、租车、自行车、公共交通、汽车或共享单车等)MAAS通过一个单一的数字平台实现搜索、预订和支付，提供门到门的无缝移动解决方案，使旅行者能够尽快到达目的地。然而，加拿大对MAAS的概念框架以及MAAS对人们出行行为的潜在影响还没有很好的阐述，目前的模型和分析方法需要改进来解释大数据下的出行决策规则。因此，该计划的长期目标是开发创新的模式、新的数据方法和适用的工具，以在加拿大建立一个高效、智能和可持续的多式联运系统。该计划的内容包括两个工作包。套餐1发展了前沿知识，以分析旅行者的旅行行为决定，考虑到人为因素(性别、年龄、地区等)的多样性。来了解他们的旅行喜好。套餐2侧重于旅行行为分析的新数学模型，以更好地预测决策。套餐2的研究结果可以为套餐1更好地理解MAAS下的多式联运决策提供支持。更具体地说，该计划的短期目标是：1)开发前沿行为模型，以了解MAAS下的消费者决策。2)发展多源数据融合技术，实现大比例尺探测器数据与小比例尺调查数据的融合。3)提出了新的特征选择方法来选择MAAS下最重要的出行预测因子。4)开发行为理论驱动的机器学习方法，将机器学习和行为模型联系起来。该计划有助于在加拿大实现可持续机动性的经济、社会和环境目标。研究结果可为政府制定长期和短期MAAS政策提供理论支持。所开发的方法论流水线对工程领域和智能交通系统的应用具有一定的参考价值，为其他研究人员的进一步发展提供了重要的基础。此外，该计划将通过将人工智能知识与交通工程领域的专业知识相结合，促进基本的跨学科进步，以培训HQP。

Mobility as a Service (MaaS) is an important means of urban transportation sustainable development that has become popular in the global transportation field recently. By integrating services of multiple providers (e.g., taxi, rental vehicle, bicycle, public transit, car- or bike-sharing, etc.) and enabling searching, booking, and payment through a single digital platform, MaaS provides door-to-door seamless mobility solutions that allows travelers to arrive destinations as quickly as possible. However, the MaaS conceptual framework and the potential influence of MaaS on people's travel behavior are not well elaborated in Canada, and current models and analysis approaches require improvement to explain the travel decision-making rules under big data. Therefore, the long-term objective of this program is to develop innovative models, new data methods, and applicable tools to generate an efficient, intelligent, and sustainable multimodal transportation system in Canada. The content of this program includes two work packages. Package 1 develops the forefront knowledge to analyze the travel behavior decisions of travelers, considering the diversity of human factors (gender, age, region, etc.) to understand their travel preferences. Package 2 focuses on new mathematical models of travel behavior analysis to better predict decisions. The research results of package 2 can provide support for package 1 to better understand the multimodal travel decisions under MaaS. More specifically, the short-term objectives for this program are to: 1) Develop forefront behavioral models to understand consumer's decision-making under MaaS. 2) Develop multi-source data fusion techniques to integrate large-scale detector data with small-scale survey data. 3) Propose new feature selection methods to select the most important predictors for travel decisions under MaaS. 4) Develop behavioral theory-driven machine learning approaches that bridge machine learning and behavior models. This program can contribute to achieving the economic, social, and environmental goals of sustainable mobility in Canada. The research results can provide theoretical support for the government to make long-term and short-term policies on MaaS. The methodological pipeline developed is valuable to engineering domains and intelligent transport system applications, which provides an important fundament to other researchers for further development. In addition, this program will promote fundamental interdisciplinary advances to train HQP through the integration of artificial intelligence knowledge with domain expertise in transportation engineering.