Date of Award:


Document Type:


Degree Name:

Master of Science (MS)


Electrical and Computer Engineering

Committee Chair(s)

Ryan Gerdes


Ryan Gerdes


Rajnikant Sharma


Tam Chantem


The highways of the world are growing more congested. People are inherently bad drivers from a safety and system reliability perspective. Self-driving cars are one solution to this problem, as automation can remove human error and react consistently to unexpected events. Automated vehicles have been touted as a potential solution to improving highway utilization and increasing the safety of people on the roads. Automated vehicles have proven to be capable of interacting safely with human drivers, but the technology is still new. This means that there are points of failure that have not been discovered yet.

The focus of this work is to provide a platform to evaluate the security and reliability of automated ground vehicles in an adversarial environment. An existing system was already in place, but it was limited to longitudinal control, relying on a steel cable to keep the vehicle on track. The upgraded platform was developed with computer vision to drive the vehicle around a track in order to facilitate an extended attack. Sensing and control methods for the platform are proposed to provide a baseline for the experimental platform.

Vehicle control depends on extensive sensor systems to determine the vehicle position relative to its surroundings. A potential attack on a vehicle could be performed by jamming the sensors necessary to reliably control the vehicle. A method to extend the sensing utility of a camera is proposed as a countermeasure against a sensor jamming attack. A monocular camera can be used to determine the bearing to a target, and this work extends the sensor capabilities to estimate the distance to the target. This provides a redundant sensor if the standard distance sensor of a vehicle is compromised by a malicious agent. For a 320×200 pixel camera, the distance estimation is accurate between 0.5 and 3 m.

One previously discovered vulnerability of automated highway systems is that vehicles can coordinate an attack to induce traffic jams and collisions. The effects of this attack on a vehicle system with mixed human and automated vehicles are analyzed. The insertion of human drivers into the system stabilizes the traffic jam at the cost of highway utilization.