<div class="section abstract">
<div class="htmlview paragraph">Autonomous ground vehicles (AGV) are comprised of a network of interconnected
components including sensors, drive-by-wire actuators, and on-board computing.
This on-vehicle network is often connected to a larger network which may include
a ground station, other autonomous systems, or remote servers. While AGV share
many features with other mobile networked devices like cell phones, the AGV
computing and networking architecture may be vulnerable in ways that other
systems are not, and the consequences of an attack may result in more severe
physical consequences. In this paper, we present a systematic study of the
network architecture of an AGV system, a cross-domain evaluation of possible
attack vectors for AGV, and an implementation of a simulated cyberphysical test
range that reveals the real-world consequences of cyberphysical attacks on
AGV.</div>
</div>