| Connected vehicles and autonomous driving systems are going to change the face of road transportation as we know it. However, we predict that the change from our present-day transportation model is going to be gradual with both human-driven and automated vehicles sharing the road throughout the process. In this work, we primarily focus on such incremental cyber transportation systems that are comprised of a mixture of human driven, partially automated and fully automated vehicles.;We have identified three crucial aspects of such cyber transportation systems that need further study: firstly, many researchers have largely ignored the human driver's role in the system. Given that the driver represents the end point in all but fully automated vehicles, it is the quality of the driver's response to the system that determines its success. Secondly, a large number of safety and information-related applications have been proposed for such systems over the years. A typical vehicle is likely to run a combination of these applications, each generating and consuming data independent of the others. Given the scarce wireless resources available to most vehicles, a system running such independent applications will scale poorly. Finally, given the fast-paced incremental nature of the system, there is a great need for a quick and cost-effective means of testing new solutions and upgrades that can help us understand their effects not only within the various domains involved (transportation, communication, driver acceptance, etc.), but more importantly, on the cyber transportation system as a whole. While such capability is not available to traditional standalone simulators, it is possible to combine multiple such simulators to model different aspects of the system. However no one has successfully developed a "complete" simulator to-date.;In this work, we take a holistic approach to cyber transportation systems and address these issues with a focus on the human driver. In particular, we begin by describing a suitable architecture that allows us to make improvements to existing applications while requiring minimal changes to their current design. We propose a high-level message fusion algorithm that can enhance the presentation of application generated information to the driver by using our knowledge of the application and also the individual controlling the vehicle. We then present low-level data fusion techniques to improve the "usefulness" of the data being exchanged between multiple applications over the wireless channel. Note that the usefulness is gauged by incorporating the human driver's perspective. We propose the architecture and working of a 3-in-1 integrated simulation tool that combines traditional traffic, driving, and network simulators (ITDNS) with additional human factors-based models to mimic human driver behavior. The Cyber Transportation Systems (CTS) Group at the University at Buffalo has already developed a preliminary version of the ITDNS with a 6 D.O.F. motion platform for driver-in-the-loop experiments, and is presently working on enhancing its capabilities. We close with an overview of selected emerging connected vehicle applications that can improve transportation efficiency, enhance safety, and boost system adoption by road users, private businesses and the government. |
