Palladyne, a U.S.-based defense and industrial technology company commercializing embedded AI and collaborative autonomy, has announced U.S. Patent 12,517,525 B1. The patent, “Path Creation, Detection and Prediction Using Primitives,” secures its proprietary Bayesian Program Learning (BPL) framework for multi-sensor target recognition, autonomous path planning, and behavioral prediction across space, air, land, and maritime environments.
According to CTO Denis Garagic, the BPL framework outperforms conventional edge AI in three critical ways: recognizing targets across sensor types without cloud reliance, executing spoken commands into robotic motion plans within seconds, and maintaining tracking during signal loss.
This patent expands Palladyne’s IP portfolio across three domains:
- Target Recognition: On-device detection and tracking across EO, IR, LiDAR, radar, acoustic, and RF sensors.
- Autonomous Path Planning: Natural-language-to-motion conversion eliminating manual reprogramming, reducing task changeover from hours to minutes.
- Behavioral Prediction: Tracking continuity through sensor dropout, occlusion, or jamming, even in communications-denied environments.
CEO Ben Wolff emphasized the strategic importance: “This patent is a direct expression of the long-term value we are building. Our framework is not incremental—it is a fundamentally different approach to machine intelligence. This IP is revolutionary and will compound in value as autonomous systems become central to defense and industrial customers accelerating adoption of edge-native autonomy.”
