Smart Mobility: Models and Applications for Smart Transit and Freight Data-Driven Informatics, Dynamic Programming, and Optimal Learning

Overview

With the compounding effects of climate change, population growth, and urbanization, traffic is becoming increasingly congested, and the effects of that congestion are increasingly worse. In particular, the rise of mega-regions due to urbanization poses a new set of problems for travelers. Many neighborhoods could benefit from use of public transit to other parts of the mega-region, but the cost of maintaining a consistently connected system increases disproportionately with the coverage area—what transportation planners call the “first–last mile problem.” For a transit agency, the first–last mile problem is challenging because it demands high flexibility under relatively low density, which implies that the cost of implementation is high.

With the promise of autonomous fleets of vehicles in the near future, the integration of real-time data with dynamic operations will become even more important. A key to effective dynamic optimization of a system (routing, scheduling, fare-setting, pre-positioning of idle vehicles, etc.) is the use of information and communications technologies (ICTs), large datasets, and urban informatics to infer outcomes in the near future. Such data could be used, for example, by a transportation network company or taxi fleet to set prices dynamically or dispatch vehicles efficiently, by bus operators to manage their vehicle arrivals dynamically in order to reduce bus bunching, or by a multimodal transportation network or a car-sharing or bike-sharing service to effectively rebalance the positioning of vehicles.

In this book, we will focus on several key elements of data-driven transport systems—what we call “smart mobility”—to address the rising challenges posed by new business models that are driven by mobile technologies. Our book will address several important issues that arise in the planning, design, and operations of modern demand-responsive transit systems. The book will make several original contributions to the advancement of knowledge and techniques in the areas of transportation systems management and operations research. Considering the latest developments in urban informatics and the increasing popularity of demand-responsive, technology-enhanced shared transportation systems (e.g., Uber), the work is well motivated and has significant potential for real-world applications. Our book will show how to create a smart and clean mobility system via several functions which are tied together as key components of a “sustainable decision support-tool” framework where dynamic operations are driven real-time information.

In this book, we show how to devise a smart decision support-tool for operations management under incomplete information. In recent years, there has been a significant increase in the number and impact of natural disasters worldwide (Center for Research on Epidemiology of Disaster, 2013). For instance, management of post-disaster debris deserves significant attention because of the various factors that attend such a hazard, including size, sorting, and disposal of the debris, as well as environmental impacts, response time, and psychological stress. Thus we will cover other types of approximation approaches, such as the partially observable Markov decision process (POMDP), Markov chains, and the hidden Markov decision process (HMDP), as well as the Markov decision process (MDP) in general.