Author ORCID Identifier


Defense Date


Document Type


Degree Name

Doctor of Philosophy


Electrical & Computer Engineering

First Advisor

Patrick J. Martin


Driving is inherently interactive, and drivers must coordinate with other vehicles, cyclists, and pedestrians to avoid collisions. Furthermore, drivers typically prefer some interaction outcomes over others. People in a rush typically cut others off and drive aggressively, while those on a leisurely outing tend to move slowly and behave more altruistically. Even as passengers in autonomous vehicles, people may exhibit these preferences. Unfortunately, autonomous vehicles do not share these preferences, instead sharing only minimal information or operating in isolation. Existing coordination and management methods do not consider these preferences when making decisions. They focus specifically on minimizing the trip duration and maximizing throughput.

Cooperative driving automation is an emerging research field in which vehicles work together to achieve individual goals. Standards organizations such as SAE International have developed a taxonomy for various collaboration categories among connected autonomous vehicles (CAVs) with varying automation capabilities. The Federal Highway Administration is developing research platforms to facilitate cooperative driving algorithm development.

This research seeks to understand how sharing passenger preferences and other high-level information affects vehicles' coordination abilities and management performance. Specifically, we attempt to answer three main research questions. 1) Can CAVs achieve better interaction outcomes by sharing high-level information, such as their passengers' preferences? 2) Can CAVs solve challenging cooperative decision-making problems by sharing high-level information, such as planned destinations? 3) How does optimizing for passenger preferences in interactive scenarios affect other traditional performance metrics?

We explore two application areas under the cooperative driving automation theme. The first scenario considers streams of autonomous vehicles simultaneously approaching an unmanaged intersection. Vehicle passengers have different preferences on how quickly they cross and in which order. Vehicles convey this information to an auction-based intersection management system installed at the intersection. The management system then assigns crossing durations and a crossing schedule to satisfy everyone's preferences as best as possible.

The second scenario investigates vehicles moving in a spatially-constrained environment such as an alleyway. In this work, we formulate a new type of finite multi-stage game we call a deadlock game. Additionally, we propose a solution method that solves general problem instances. This work provides the foundation for continued research into equilibria refinement and satisfying passengers' preferences on the outcomes.


© The Author

Is Part Of

VCU University Archives

Is Part Of

VCU Theses and Dissertations

Date of Submission


Available for download on Thursday, April 25, 2024