On-board Telemetry Monitoring in Autonomous Satellites: Challenges and Opportunities

TL;DR

This paper presents a framework for explainable neural fault detection in satellite telemetry, enabling interpretable anomaly identification with minimal additional computational load.

Contribution

It introduces peepholes, a method to derive semantically meaningful encodings from neural activations, improving interpretability for onboard satellite fault detection.

Findings

Peepholes enable semantic characterization of anomalies.

The framework supports anomaly localization in reaction-wheel telemetry.

Minimal computational overhead makes it feasible for onboard deployment.

Abstract

The increasing autonomy of spacecraft demands fault-detection systems that are both reliable and explainable. This work addresses eXplainable Artificial Intelligence for onboard Fault Detection, Isolation and Recovery within the Attitude and Orbit Control Subsystem by introducing a framework that enhances interpretability in neural anomaly detectors. We propose a method to derive low-dimensional, semantically annotated encodings from intermediate neural activations, called peepholes. Applied to a convolutional autoencoder, the framework produces interpretable indicators that enable the identification and localization of anomalies in reaction-wheel telemetry. Peepholes analysis further reveals bias detection and supports fault localization. The proposed framework enables the semantic characterization of detected anomalies while requiring only a marginal increase in computational resources, thus supporting its feasibility for on-board deployment.