Advanced data-driven optimization tools for public transit scheduling and parcel delivery

点击翻译按钮获取中文摘要

GIRO Inc. is a world leader in the development and commercialization of optimization-based softwares for planning the operations of public transit and postal companies. This research project aims at developing new data-driven optimization algorithms for GIRO's Hastus and GeoRoute softwares dedicated to public transport and postal companies, respectively. Some of the research results will be applicable to other contexts. For public transit companies, we propose to develop state-of-the-art solution algorithms for two problems: Integrated bus and duty scheduling problem with trip shifting that allows to slightly change the schedule of certain bus trip; and Electric bus scheduling which requires planning battery recharges. In postal companies, we observe an increase in the number of parcels to deliver with the advent of e-commerce. The postmen are assigned to compact territories and the average workload must be balanced as much as possible between these territories. In this project, we will conceive an algorithm to rebalance the workloads on a given day when some postmen are overloaded. As generic developments, we plan to improve the optimization algorithms used in GIRO's softwares and in many other companies by integrating machine learning (artificial intelligence) tools into them. These tools will exploit a huge amount of data that have become available via, e.g., mobile sensors. This research will yield several benefits to Canada. First, a Canadian-owned private company, GIRO, will enhance two of its softwares that are commercialized worldwide. Second, Canada Post and several Canadian public transport companies that plan their operations with either GeoRoute or Hastus will have access to improved algorithms that will help them realized savings. Finally, five graduate students and two postdoctoral fellows whose specialization is often sought in optimization-based companies in Montreal and elsewhere in Canada will be trained.