SBIR/STTR Award attributes
To be mission-relevant, Navy unmanned surface vehicles (USVs) must perceive and behave like manned ships. Cameras are valuable for supporting human-like perception because their high-resolution detail enables ship classification, tracking, COLREGS determination, and threat characterization. Human watchstanders use a variety of cues and heuristics to scan the surroundings and contacts optimally, but current camera systems do not offer the same intelligent, adaptive perception. To deliver capable, adaptable maritime perception, we propose to design and demonstrate the feasibility of Adaptive Perception Behaviors (APB) for smart maritime cameras. APB enables cameras to behave like human watchstanders, without intervention from external autonomy systems, by intelligently balancing competing perception behaviors, such as examining contacts and scanning the surface for new contacts. APB couples our SeaSaw smart maritime camera with a pan-tilt mount and intelligent behavior optimization to create a self-controlled camera that can detect, classify, and track ships and decide on the best order in which to inspect surface contacts to ensure safe navigation. Beyond simple surface situational awareness, we are augmenting APB with enhanced detection for safe navigation, false track rejection, AIS spoofing detection, zoom control, and stabilization. We will also install APB on alternative platforms and camera systems. APB will be demonstrated at ANTX and other events to collect mission-relevant feedback. By decoupling autonomy and perception, APB simplifies USV design and facilitates perception hardware and behavior upgrades on USVs, making USVs more capable and adaptable.
