面向时空随机请求的群智物流关键技术研究

High-efficient city-wide logistics plays a fundamental but significant role in shaping the Smart City and Industry 4.0. To speed up the shipping process in the city, more personnel and dedicated vehicles are usually employed, as a result, more cost and serious environmental problems are generated. With the rapid development of mobile internet and global positioning technologies, and the proliferation of vehicles (e.g., taxis, buses, private cars) in everyday use, a novel city-wide package-delivery paradigm, Crowdsourcing Logistics (CSL), which accomplishes package deliveries by leveraging the extra loading capacity provided by other on-road vehicles (i.e. not the dedicated vehicles for the purpose of goods transportation), is gradually becoming realistic. We name the extra loading capacity provided by other on-road vehicles as “crowdsourced resources”. However, the stochastic characteristics of crowdsourced resources and package-delivery requests in space and time bring us the challenges in matching, modelling and optimization of the crowdsourced resources, which makes the building of an efficient crowdsourcing logistics system rather difficult. In this project, we intend to address the above challenges through investigating the following three main research issues in CSL: (1) the optimization of nodes (i.e. the determination of locations and numbers of nodes), with the objective of improving the matching ratio, (2) the modelling of package routing time on edges, with the objective of measuring and comparing different package routing-paths, and (3) the package routing-path discovery, with the objective of optimizing the use of crowdsourced resources in CSL network. This project aims to develop key techniques for CSL, laying necessary theoretical foundation for the design and analysis of economic, high-efficient, and environmentally-friendly CSL systems.

高效的城市物流是推进智慧城市与工业4.0建设的重要基础。为提高城市物流效率，目前主要采取重资源投入的手段，存在成本高、环境不可持续等问题。基于移动互联网、全球定位等技术的日趋成熟及移动载体（出租车、公交车、私家车等）的广泛普及，本项目提出一种新型的物流模式—群智物流，即：利用非物流专用移动载体在响应自身运输任务时产生的富余运力（统称为群智资源）辅助物流运输。但物流请求和群智资源具有时空随机性，为构建高效的群智物流系统带来“难匹配”、“难建模”、“难优化”三大挑战。基于此，本项目将重点研究：（1） 群智物流网络节点优化方法，以提高群智资源的匹配率;（2）群智物流网络边流通时间建模方法，以衡量和比较不同物流路线的流通时间；（3）群智物流路线优化方法，以动态调度群智资源并自适应发现最优物流路线。项目预期形成群智物流系统关键技术框架，为实现低成本、高效率和环境可持续的城市物流提供理论和技术支撑。

群智物流旨在利用非物流专用移动载体在响应自身运输任务时产生的富余运力（统称为群智资源）辅助城市物流运输。为了克服群智资源“难匹配”、“难建模”、“难优化”三大难点问题，本项目研究了：1）群智物流网络节点优化方法，从给定的候选中转站集合中选取最优子集，并借助中转站，采用间接匹配的思想，达到了提高群智资源的匹配率的目的；2）群智物流网络边流通时间建模方法，挖掘载客出租车 GPS 轨迹，通过时空聚类方法，并基于聚类结果深刻理解出租车载客行为的时空特性及呈现的模式，从而实现城市内两点间的等待载客事件时间及行驶时间的精确建模；3）不同应用场景下，针对不同优化目标（包括最大化包裹准时到达概率、最小化包裹运输时间、最小化额外行驶距离等），分别提出了群智物流路线优化方法。在项目执行期间，共发表学术论文31篇，其中SCI期刊源论文19篇，ACM/IEEE会刊论文9篇，JCR1区论文11篇。CCF-A类会议论文2篇，CCF-B类会议论文1篇。1篇论文入选ESI高被引论文。申请发明专利7项。培养博士研究生3名，硕士研究生7名，其中1名硕士荣获校级优秀硕士学位论文。项目研究成果较大程度上超过了预期成果。特别地，以最小化包裹运输时间的群智物流路线优化方法的研究，受到了国际IEEE Spectrum、国内长沙晚报、新浪科技等网站的长篇幅报道。其研究思路受到了国际同行的普遍认可。同时，项目执行期间，举办了多个相关学术论坛活动，包括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，项目形成了群智物流系统关键技术框架，为实现低成本、高效率和环境可持续城市物流的实际落地和应用提供了较强的理论和技术支撑。

内容获取失败，请点击重试