Palladyne AI (NASDAQ: PDYN) announced the issuance of U.S. Patent 12,517,525 B1 for its Bayesian Program Learning framework that enables target recognition, autonomous path planning, and behavioral prediction across multiple sensor types.

The patent, titled "Path Creation, Detection and Prediction Using Primitives," covers technology that operates across space, air, land, and maritime environments. The framework processes data from electro-optical, infrared, LiDAR, radar, acoustic, and radio frequency sensors without requiring cloud connectivity.

According to the company's press release, the patented technology includes three capabilities: detecting and tracking moving targets across multiple sensor types on-device, converting natural-language commands into motion plans, and predicting target behavior during sensor disruptions or jamming.

"We believe our patented BPL framework does three things no conventional AI system can match at the edge: it recognizes targets across multiple sensor types without the cloud, it turns a spoken instruction into an optimized robotic motion plan in seconds, and it keeps tracking even when the signal goes dark," said Denis Garagic, the company's chief technology officer and co-founder.

The Salt Lake City-based defense and industrial technology company develops AI systems for autonomous robotics, surveillance, and edge computing applications. This represents the company's second patent issued in four months, expanding its intellectual property portfolio in the autonomous systems sector.

Palladyne AI focuses on embedded artificial intelligence and collaborative autonomy solutions for defense and industrial markets, with emphasis on American-developed platforms designed for government and public-sector customers.