以乘客为中心的智能移动出行关键技术研究

The smartness of people transportation is a key indicator of the development of smart cities. The increasing popularity of smart phones and other mobile devices has made the real-time location of passengers and vehicles easily available. In this context, a new transportation mode relying on the mobile devices is gradually becoming dominant. We name the new mode as mobile transportation. Mobile transportation based on taxis is the most popular and effective in matching taxi demands and supplies in real time, however, it is still far from smart. To make the matter worse, it fails to meet the huge transportation demands due to the lack of taxi vehicular resource. Thanks to rich delivery resources resulted by the huge number and rapid growth of private cars, mobile transportation based on ridesourcing cars is thus recognized as a new solution, however, the newly introduced “dynamic price” also creates new research challenges. In this project, based on the big data collected when passengers interacting with physical and cyber spaces, we intend to build a smart passenger-centric mobile transportation system to address the new challenges. More specifically, focusing on the new feature of dynamic pricing, we intend to investigate the following three main research issues: (1) the modelling passengers’ reaction to dynamic prices, with the objective of building the mathematical relationship among the passengers’ reaction, dynamic prices and the demand elasticity, (2) the prediction of dynamic prices, with the objective of predicting dynamic prices for passengers accurately, (3) the real-time inference of trip purposes and its validation, with the objective of inferring the trip purposes in real time. This project aims to develop key techniques for mobile transportation systems based on ridesourcing cars for passengers, laying necessary theoretical foundation for the design and build of the future transportation ecosystem.

居民出行的智能化程度是衡量智慧城市发展水平的重要性能指标。智能手机等移动终端的普及使得乘客和车辆的实时位置、状态等信息容易获得，催生并使基于移动端的出行模式—移动出行成为新趋势。基于出租车移动出行实现了乘车需求与出租车载客资源的实时匹配，但仍无法满足大量乘车需求。基于专车移动出行因具备更加广泛的载客资源，为满足大量乘车需求提供了新途径，但新的动态价格特征也为智能化带来了新挑战。本项目聚焦于动态价格，基于用户与物理信息二元空间交互产生的大数据，以乘客为中心，为实现移动出行智能化研究三方面关键技术：1、乘客反应行为建模方法，以刻画乘客对动态价格的反应行为、动态价格、出行需求弹性之间的关系；2、动态价格预测方法，以高精度预测动态价格；3、个人出行意图实时推测及验证方法，以及时准确推测乘客乘车意图。项目预期形成乘客智能移动出行的系统理论和关键技术框架，也将为构建未来移动出行生态系统提供理论支撑。

居民出行的智能化程度是衡量智慧城市发展水平的重要性能指标。智能手机等移动终端的普及使得乘客和车辆的实时位置、状态等信息容易获得，催生并使基于移动端的出行模式—移动出行成为新趋势。基于出租车移动出行实现了乘车需求与出租车载客资源的实时匹配，但仍无法满足大量乘车需求。基于专车移动出行因具备更加广泛的载客资源，为满足大量乘车需求提供了新途径，但新的动态价格特征也为智能化带来了新挑战。本项目聚焦于动态价格，基于用户与物理信息二元空间交互产生的大数据，以乘客为中心，主要解决以下三大关键性挑战：1) 人类移动的演化在时间和空间上存在多尺度、多依赖的复杂高阶关联，难以全面有效地建模；2) 普适场景中安全可用的信息来源有限，难以准确地刻画深层次细粒度的移动语义；3) 人类行为语义普遍存在标签稀疏性问题，难以有效地训练高性能模型。在项目执行期间，共发表论文37篇，其中SCI期刊源论文23篇，ACM/IEEE会刊论文9篇，JCR-1区论文16篇，CCF-A类论文3篇，CCF-B类论文8篇。申请发明专利14项。培养博士研究生6名，硕士研究生16名。项目研究成果较大程度上超过了预期成果。同时，项目执行期间，举办了多个相关学术论坛活动，包括HHME人机和谐环境学术会议、CCF城市计算与智能感知论坛、西部地区青年学者智慧计算论坛等 。综上，项目形成了智能移动出行关键技术框架，为实现高效率、低成本、环境可持续智能移动出行的实际落地和应用提供了较强的理论和技术支撑。