CAREER: Pathway to a Driverless Highway Transportation System: A Behavior Analysis and Trajectory Control Approach

University Of South Florida, Tampa FL

Investigators

Abstract

This Faculty Early Career Development (CAREER) Program grant investigates near-future highway traffic with a portion of vehicles being automated and controllable by roadside units (e.g., using driverless car and connected vehicle technologies). This research project explores strategies of controlling the trajectories of these automated vehicles to not only optimize their own performance but also improve the experience of nearby human drivers. These novel trajectory-based control concepts and models can be applied to related industrial developments in highway infrastructure, electronic devices, and vehicle technologies. The research outcomes expect to provide a methodological foundation for evaluating the feasibility and the potential benefits of transferring the current human-driven highway transportation system into one that is fully automated. Further, the research outcomes are to be integrated into creative education initiatives to motivate and prepare students to pursue an engineering career. This project establishes a virtual traffic laboratory where students can play interactive driving games during classroom learning. This experience-based virtual learning platform supports engineering curriculum developments, outreach to high school students and underrepresented groups, and workshops with industrial collaborators and academic peers. The research objectives are to: (i) discover fundamental patterns of interactions among neighboring drivers on a highway and (ii) create strategies for controlling trajectories of distributed automated vehicles to improve the overall traffic performance based on these interaction patterns. This research establishes a virtual highway experiment platform by integrating high-fidelity driving simulators with off-the-shelf traffic simulators, resulting in a flexible highway driving environment that facilitates the easy recording of driver-to-vehicle interactions. Further, this effort creates a new data analysis method that decomposes collected trajectories into a set of elemental segments, allowing us to apply data mining techniques to uncover driver interaction patterns. These efforts aim to overcome two prominent challenges in transportation research: the difficulty of collecting high-fidelity vehicle trajectory data and the lack of quantitative methods for analyzing such data. Based on these discoveries, a traffic control framework is to be developed that plans trajectories for driverless vehicles in order to improve traffic efficiency, reduces environmental impacts, increases safety, and improves the experience of all drivers. This research advances the scope of trajectory optimization methodologies from traditional isolated individual trajectories to multiple interdependent and interactive trajectories in a shared transportation channel.