JAMES GOPPERT

 

Research Associate Professor Candidate

 

ABSTRACT

 

Unmanned aerial systems (UASs), also known colloquially as drones, are proliferating due to their efficiency and the minimal risk to human life when they conduct tasks such as surveillance, inspection, and search and rescue. NASA has envisioned an Advanced Air Mobility (AAM) system leveraging UASs that will be in place by 2030 where the movement of people and goods will be brought off of the ground and into the sky. To reach this vision, we must increase the safety and security of UAS operations. Increasing trust in aerial autonomy requires careful analysis of the coupling of physical components with digital control programs. Hybrid model checking techniques have been developed to consider this coupling, but the nonlinearities of the system dynamics and complexity of the control programs makes this difficult to realize for non-trivial systems. In addition, an intelligent attacker may leverage information about a UAS to construct attacks which expose system vulnerabilities not seen during normal operation. This talk will discuss approaches to address the fundamental trust gap for the AAM vision of 2030. First, we will discuss methods for Cyber-Physical Vulnerability Analysis (CPVA) including a mixed-reality sensor emulation framework. Next, correct-by-construction autopilot design will be discussed, which mitigates the complexities of performing software verification and validation (V&V) by constructing the code to have provable safety properties. We leverage Lie group theory to construct control and estimation algorithms with more efficient and less conservative reachable sets for proving safety under bounded disturbances. Finally, we will discuss counter-UAS system development for tracking nefarious actors with a low-cost self-calibrating camera network and a trajectory planning algorithm to detect weaknesses in the system.