CHANTILLY, Va. – The Aerospace Corporation and Google Public Sector, both based in Virginia, are collaborating on an artificial intelligence (AI)-based system designed to help satellite operators detect and investigate spacecraft anomalies.
The organizations have developed a proof-of-concept tool that applies agentic AI to satellite operations. The goal is to reduce the time engineers spend monitoring spacecraft health and investigating unexpected behavior. The effort focuses on proliferated low-Earth orbit (pLEO) constellations, where operators may be responsible for monitoring hundreds or even thousands of satellites simultaneously.
Related: Aerospace, Google apply AI to satellite anomaly detection for LEO constellations
Satellite operations generate more data than ever
As satellite constellations expand, operators face an increasingly difficult challenge: turning large volumes of telemetry into actionable information.
Every spacecraft continuously generates health and status data covering power systems, propulsion, payloads, thermal performance, and other onboard subsystems. Engineers often review that data alongside information from ground stations and environmental conditions when investigating anomalies.
As fleets grow, manually correlating those information sources becomes more difficult. Important warning signs can be buried within thousands of routine telemetry updates, making it harder for operators to identify developing problems before they affect mission performance.
The Aerospace Corporation said its collaboration with Google Public Sector aims to simplify that process by consolidating multiple telemetry streams into a single operational view. The AI also detects unusual system behavior automatically. Rather than relying only on predefined alarm thresholds, the tool is designed to recognize behavioral patterns that may indicate an emerging anomaly.
AI supports engineers during anomaly investigations
Satellite anomaly resolution traditionally requires engineers to compare information from multiple software applications while uncovering whether an unexpected condition represents a true system problem or a temporary operational event.
The proof-of-concept tool minimizes that workload by correlating telemetry, payload status, and ground network information within a single interface. According to the organizations, the system can identify relationships between different data sources and provide engineers with additional context during an investigation.
The companies said the objective is to help engineers reach root causes more quickly rather than replace human decision-making. Operators remain responsible for evaluating recommendations and determining the appropriate response to any spacecraft anomaly.
Agentic AI expands operational decision support
The collaboration also highlights how AI is beginning to play a larger role in satellite operations.
Earlier AI applications often focused on analyzing historical mission data or automating routine tasks. Agentic AI introduces a different approach by letting software identify potential issues and assist with more complex investigative workflows.
For satellite operators, that capability could become more valuable as commercial and government organizations deploy larger constellations supporting Earth observation, navigation, and national security missions. Larger fleets create more potential fault conditions and force operators to make more decisions in less time.
AI-assisted monitoring may help operators detect subtle changes that conventional threshold-based monitoring systems could overlook. At the same time, the system could allow engineering teams to focus their attention on higher-priority technical issues.
The proof-of-concept remains an early demonstration, but it illustrates how AI is evolving beyond routine automation to support operational decision-making across complex satellite fleets.
