FDIR for Autonomous Space Systems for Anomalies and Cyberattacks

Abstract

Verbatim from NASA NTRS. Not paraphrased, not summarized.

Fault detection, identification, and remediation (FDIR) is a challenging task for any space system. Faults are not necessarily binary in nature, and several different faults may all create the same symptoms or bad performance. This is compounded when the consideration of cyberattacks is likewise added, and further made daunting when the system in question is intended to perform its tasks autonomously. In this work, a spacecraft in the tug-cargo configuration is simulated during an RPOD maneuver. During this maneuver, several zero-dynamics attacks are simulated that attack the spacecraft’s sensors, all of which first cause the spacecraft either fail catastrophically, or initiate an auto-abort. Then, a discriminative unscented Kalman filter previously used to identify unknown dynamics is used instead to identify faulty or misleading sensors, and successfully perform the docking maneuver in spite of these failures. The simulated spacecraft is simulated in multi body problem using the rigid body formulism on special Euclidean group SE(3). Additionally, Morse-Lyapunov control is applied considering constrained, distributed control among reaction control system thrusters with saturation limits and duty cycle constraints.

Authors

• Matthew M Wittal Kennedy Space Center
• Michael Czernec Aegis Aerospace (United States)

Keywords

• Fault
• FDIR
• discriminative
• Kalman
• autonomy
• GN&C