Using Fault Detection/Identification Software with Local Awareness Sensors to Improve Satellite Resiliency

With 20,000 trackable orbital objects and hundreds of thousands of orbital debris too small to be cataloged, satellites today operate in an increasingly congested environment as more and more launches are performed. It is also common for a satellite to not be in 24/7 contact with the ground, this break in the communication network implying that when there is an anomalous event, there is often a longer downtime than desired. Coupling a probabilistic Artificial Intelligence (AI) system with local awareness sensors running onboard the vehicle, will enable the satellite to execute maneuvers to reduce the chances of an on-orbit collision and have higher resilience to the effects and dangers of space. To reduce the chances of the satellite being struck by debris the autonomous system could use local cameras to plan and execute maneuvers around reducing the chance of an on-orbit collision. By monitoring the local structural state health of the satellite using sensors, the satellite could be alerted when it has been struck by debris and which area of the satellite has been effected. The information from the monitoring local structure state health system can be used by an onboard autonomous system to not only update the operator but also plan around the damaged area if possible, so the overall mission can still be accomplished. At the Space Vehicle Directorate in the Air Force Research Laboratory (AFRL), software has been built to investigate aspects of this concept. This software has been designed to be modular with each executable performing a specific function. Each executable falls into one of three different categories: processing sensor data, analysis of the processed data, and decision making using the analyzed information (data fusion). Probabilistic model are used to infer and recommend a proper course of action. The initial capabilities of this concept have been tested with a hardware-in-the-loop (HIL) testbed. This testbed allows us to work with realistic attitude information, simulated high-fidelity orbital dynamics, and hardware emulating on-board local awareness sensors local awareness sensor hardware. Future work will focus on taking a version of this software and implement it on repeated 12-27U CubeSat missions, expanding our software and hardware library and improving the resiliency of the satellite.