U

Autonomous Spacecraft Subsystem Fault Detection, Isolation, Diagnosis, and Recovery

Stottler Henke has developed a sophisticated spacecraft subsystem Fault Detection, Isolation, and Diagnosis capability combined with our Aurora Space Application Planning and Scheduling framework.  Together these provide a high-level, closed loop system for fully automated detection of faults and anomalies; diagnosing the underlying causes; and planning, scheduling, and executing recovery and reconfiguration activities. Following a detected fault, this automated capability helps to maximally recoup spacecraft subsystem functionality, and optimally fulfill mission objectives to the degree possible.  

Our MAESTRO (Management through intelligent, AdaptivE, autonomouS, faulT identification and diagnosis, Reconfiguration/replanning/rescheduling Optimization) architecture is designed for straight-forward reapplication to different spacecraft subsystem management problems.  

MAESTRO can be configured with standalone subsystem or problem-specific modules, or they can be easily integrated with one another to execute as a closed-loop in a variety of computational environments, including in highly distributed situations. We integrated our AI Modules within NASA’s core Flight System (cFS) so that they can be used (through cFS) on a wide variety of spacecraft, from large human exploration vehicles to small scientific instruments down to the cubesat level.  We also integrated these same AI Modules on Montana State University’s RadPC (radiation-tolerant processing FPGAs) in an experiment onboard the ISS that detected and diagnosed issues with the EPS and RadPC itself.  Other applications include the xEMU Portable Life Support System (PLSS), Gateway PPE Electrical Power System (EPS), and Mars Transit Vehicle.

During normal operations, MAESTRO monitors onboard sensor values in order to automatically characterize subsystem components in preparation for detecting failures.  Based on that characterization, it automatically predicts resource availability over time and schedules the actions (i.e., determining what activities will occur and when, along with the modes of the associated equipment). During a failure scenario, MAESTRO follows an operational sequence to:

  • Detect the problem
  • Immediately safe the spacecraft to minimize damage
  • Diagnose the problem and determine the root cause
  • Determine potential feasible courses of action (COAs) given the failed components or set of possible failed components
  • Determine the impact and ramifications of each COA
  • Select the most appropriate COA
  • Generate the detailed schedule/sequence of actions to implement the COA
  • Adaptively execute the required actions.

MAESTRO has already diagnosed real known and novel faults in various spacecraft hardware subsystems and been integrated with NASA’s core Flight System (cFS) and NASA JSC’s IRIS architecture.  MAESTRO’s symbolic Model-Based Reasoning (MBR) techniques are being combined with our ADTM SOM ML technology to create a hybrid system that combines the advantages of each approach while mitigating each’s weaknesses.