Richard Stottler, Stottler Henke Associates Inc.; Abhimanyu Singhal, Stottler Henke Associates Inc.; Christopher Healy, Stottler Henke Associates Inc.; Sowmya Ramachandran, Stottler Henke Associates Inc.; Kerry Quinn, Astrobotic; Joseph Palmieri, Astrobotic; Shawn Logan, Astrobotic
Keywords: anomaly detection, fault management, space situational awareness, machine learning, model-based reasoning, artificial intelligence
Abstract:
An important component of Space Situational Awareness (SSA) / Space Domain Awareness (SDA) is knowledge of the true status of friendly assets and whether any assets are under attack. Therefore, it is important to be able to detect faults and other anomalies, and determine the components involved and the root cause as well as whether that root cause is likely an external attack. During space conflict, communications to satellites may be disrupted, requiring them to intelligently and autonomously “take care of themselves,” i.e., effectively detect faults, diagnose their root causes, and develop and execute recovery plans, autonomously, without necessarily being able to communicate with ground controllers. This lack of communication is analogous to lunar rovers and power systems where communication can be disrupted by terrain and other factors.
Astrobotic, for NASA, is developing a rover that traverses over the lunar surface to an advantageous position, then unfurls a 60’ high photovoltaic mast to provide power for other lunar systems. Astrobotic’s Vertical Solar Array Technology (VSAT) will egress from its lander, transit to the desired location (near the lunar South Pole), “wiggle” into the lunar soil, and then deploy the 60’ high Roll Out Solar Array (ROSA) to generate and then distribute power to other lunar systems. The VSAT will include several subsystems, such as mobility, internal and external (to provide power to external systems) electrical power systems, thermal management, and array deployment, each of which must work smoothly in order for the operation to succeed. As the VSAT moves around the surface of the Moon, sensors are constantly providing information on how much traction is available and how quickly the rover is moving. As the solar array is unfurled, a gimbal system and inertial measurement units (IMUs) continuously monitor the array’s movement, including any lean. If the array leans too much, the solar array can buckle—worse, the entire rover may be at risk of tipping over, failing the mission. Since the array is so tall compared to VSAT’s wheelbase, even just a few degrees of lean would be disastrous. This situation may be very dynamic, denying ground controllers enough time to correct any problem, given the round trip communication delays.
It is therefore important that the VSAT be equipped with the means to quickly detect problems, perform diagnosis and root cause determination, and quickly safe the system. Traditionally, Fault Detection, Isolation, and Recovery (FDIR) systems have utilized Model Based Reasoning (MBR), which requires knowledge of the subsystem design and the behavior of components down to the desired level of diagnosis. To the degree this information is readily available, it is important to make good use of it. However, the field of machine learning (ML) has shown that systems can also learn, offline, the normal behavior of complex systems in many different environments and states, and then detect abnormal behavior in real time. These systems can also be trained with known abnormal states, and recognize these more specifically when they occur.
With the new types of subsystems (such as mechanical components and related sensors) came new challenges to be overcome. Some concerns included quick reaction times needed to avoid tipping or buckling during mast deployment and, at the opposite end of the spectrum, detecting very gradual changes, hard to discern in sensor noise (the mast moves very, very slowly while tracking the Sun). In some cases, data is severely limited, reducing the applicability of a pure ML approach.
In our previous work, we outlined our modular approach to fault detection and diagnosis utilizing MBR and ML as well as a third independent method called the Thermodynamic Reasoning and Intelligent Anomaly Detection (TRIAD) system. Similarly to how aggregate variables such as thermodynamic variables such as pressure and temperature can summarize microstate variables (e.g. the speeds of individual molecules), TRIAD utilizes aggregate quantities such as mean, minimum, maximum, and Fourier Transforms to characterize anomalies. We also described how this hybridization enables additional confidence in diagnosis, as the advantages of each approach are emphasized while the disadvantages are mitigated, and summarized how we planned to apply these methods on the VSAT platform and subsystems.
This paper will describe progress on this work since our last paper, presented at AMOS 2023. This includes validation of the hybrid approach to fault detection, diagnosis, and recovery via a physical simulation of the VSAT platform as well as results from multiple fault detection modules. We enumerate many relevant scenarios, developed in conjunction with Astrobotic to best capture realistic faults, as well as metrics from our approach.
We show that the previously discussed methods are capable of both detecting and characterizing mechanical anomalies from simulated VSAT telemetry data within tens of milliseconds of the faults occurring, well below the allotted “reaction time” of 100 milliseconds. The paper will present quantitative results for a large range of fault scenarios, including soil collapse, soil slippage, ROSA levelling errors, and a wide variety of sensor faults. Both MBR and TRIAD were effective at detection and diagnosis and, as mentioned in our previous work, we identified several areas where hybridization of both techniques provides a significant advantage over the use of just one or the other. We conclude with a discussion on the direction this work will take in the future. Based on these results, Astrobotic plans to include MAIFLOWER on the actual lunar VSAT, with integration beginning this Fall.
Date of Conference: September 17-20, 2024
Track: Space-Based Assets