Robust State Estimation in Multi-Agent Systems using Pairwise Measurements
Multi-Agent Systems (MAS) such as fleets of autonomous vehicles have a wide range of applications, ranging from search-and-rescue operations to commercial package delivery. They use a combination of onboard sensors and communications to fulfill their tasks in a safe and coordinated manner. Consequently, adversaries seeking to compromise the performance or safety of the MAS and/or the surrounding human infrastructure can do so using cyberattacks injected through these sensing and communication channels. In this project, we look at various techniques to reinforce the cybersecurity of an MAS by exploiting the pairwise measurements between agents, for e.g., cameras pointed from one agent to another. Special emphasis is placed on fast, scalable algorithms which are suitable for large-scale MAS applications.