PiLogic partners with Air Force lab to test satellite fault-prediction software

WASHINGTON — PiLogic, a startup developing artificial intelligence software to identify faults and predict failures in satellites, has signed an agreement with the U.S. Air Force Research Laboratory to test whether its technology can improve the way spacecraft operators detect and diagnose problems in orbit.

The Los Angeles-based company said it will work with AFRL under a two-year Cooperative Research and Development Agreement, or CRADA, focused on spacecraft electrical and power systems. Engineers will use an AFRL cubeSat experiment launched in 2022 through the Defense Department’s Space Test Program as a platform to evaluate PiLogic’s software.

The company’s software is designed to analyze onboard sensor data, detect anomalies, predict potential failure modes and recommend corrective actions. PiLogic argues that its approach, based on probabilistic reasoning and automated causal analysis, could be used as an alternative to traditional monitoring systems that rely on predetermined rules and thresholds.

Satellite operators typically use rules-based systems that trigger alerts when specific parameters move outside prescribed limits, such as battery voltage dropping below a threshold or temperatures exceeding allowable ranges. PiLogic contends that such systems can struggle when multiple symptoms occur simultaneously or when the source of a problem is ambiguous.

Instead of evaluating individual conditions in isolation, the company’s software attempts to determine the most likely underlying cause by combining engineering models, physics-based relationships and probability theory.

Johannes Waldstein, PiLogic’s co-founder and chief executive, said the technology is intended to provide engineers with an explainable understanding of spacecraft behavior rather than simply generating alerts.

AFRL plans to use the project to explore new approaches to spacecraft autonomy and health monitoring.

“We are excited to evaluate the next generation of autonomy for satellite health monitoring with true causal understanding,” said Joseph Melville, satellite autonomy lead at AFRL.

The agreement comes as defense agencies and aerospace companies are examining how artificial intelligence can be applied to military space operations. Much of the investment in AI has focused on large language models and machine-learning systems trained on vast datasets. PiLogic is pursuing a different approach.

The company describes its technology as “exact AI,” emphasizing probabilistic reasoning rather than generative AI. Waldstein said many aerospace and defense applications lack the l