HONOLULU - Lockheed Martin and the U.S. Navy demonstrated an artificial intelligence and machine-learning capability designed to rapidly update antisubmarine warfare (ASW) acoustic classification models during the Rim of the Pacific (RIMPAC) 2026 exercise.
The SensorMAX system monitored subsurface acoustic data collected from sonobuoys deployed by MH-60R helicopters and enabled operators to incorporate newly identified acoustic signatures into updated artificial intelligence (AI) classifiers before transmitting the updated models back to deployed aircraft, Lockheed Martin says.
The demonstration highlights an emerging challenge for military artificial intelligence: maintaining model relevance as operational environments change while enabling updates in contested environments with limited connectivity and computing resources.
Related: Special Operations Command asks industry for uncrewed systems and data processing for special forces uses
Excercise details
During the exercise, MH-60R helicopters equipped with SensorMAX conducted ASW missions using sonobuoys as underwater acoustic sensors. The system processed acoustic data from multiple sensor feeds, assisted aircrews with target identification, and enabled operators to update AI classification models during the mission, according to the company.
Lockheed Martin says the workflow consists of four steps: capture, label, update, and deploy. When operators identify a new acoustic signature, they can label the relevant data, update the AI classifier, and transmit the updated classifier package to aircraft using an encrypted data link. The company says classifier updates can be completed in less than five minutes per iteration.
The capability is designed to address a key challenge associated with military AI adoption: ensuring that deployed systems continue to provide relevant analysis as mission environments evolve. Acoustic signatures can vary based on ocean conditions, platform characteristics, sensor performance, and other factors, requiring AI systems to adapt as new data becomes available.
To the MAX
SensorMAX is built on Lockheed Martin’s Spectral Foundation Model, which the company describes as an AI foundation model designed to interpret spectral energy data from sensors. The company says the technology can support multiple sensor types, including acoustic and electromagnetic systems.
Related: Lockheed Martin receives $44.6 million Navy order for submarine EW modernization
Lockheed Martin also highlights the system’s ability to support AI processing and model updates at the tactical edge without relying on continuous access to centralized data centers. The company says SensorMAX uses compact classifier models that can be distributed through secure software updates, enabling AI capabilities to be updated across deployed platforms.
The demonstration reflects broader Department of Defense efforts to integrate artificial intelligence into military systems while maintaining cybersecurity, data provenance, and operator oversight. Rather than replacing operators, SensorMAX is designed to provide AI-assisted analysis that helps aircrews process larger volumes of sensor data during missions.
"Today’s battlespace requires the defense industry to accelerate capability delivery and seamlessly integrate commercial technology," said Devon Rodgers, vice president and general manager of Undersea Mission Systems at Lockheed Martin. "At RIMPAC 2026 our mission-focused engineers demonstrated exactly how we are meeting that challenge."
Lockheed Martin developed SensorMAX as a family of AI and machine-learning applications intended for sensor processing and classification. The company says future applications could extend beyond acoustic sensing to other systems that collect spectral data.
