Perception Autonomy for Vessel Escorts (PAVE)

Ship traffic in harbors is increasing, and asymmetric threats from small craft are also proliferating as adversaries seek methods to inflict costly damage on US assets with minimal resources. Escort details provide port and harbor security to combat these threats, but these details have significant crew requirements. Automating the escort mission with uncrewed surface vessels (USVs) would reduce risk and personnel requirements; however, the Navy’s autonomy and perception efforts have largely focused on open ocean scenarios, whereas in harbors, detection ranges are shorter, backgrounds are cluttered, and obstacles are more numerous. To address the perception challenges associated with in-harbor operations, we propose Perception Autonomy for Vessel Escorts (PAVE). PAVE is an extensible perception software system for USVs that detects, localizes, and tracks nearby vessels to enable interdiction maneuvers if necessary. PAVE perception processes electro-optical (EO) and infrared (IR) imagery to effectively detect, classify, localize, and track surface contacts, day or night, in cluttered backgrounds. We will develop enhanced monocular depth estimation for improved target localization, automatic threat assessment for asset prioritization, intelligent perception optimization to enhance system confidence and accuracy, and extensions to our existing C2 user interface for situation awareness and multi-vehicle coordination to facilitate escort missions.