Supervised Autonomy for Advanced Perception and Hydrocarbon Leak Detection

Service providers for subsea inspection, maintenance, and repair (IMR) generally utilize remotely operated vehicles (ROVs) to evaluate and manipulate underwater infrastructure. The cost of these operations can be considerable, mostly due to the need to deploy a large surface vessel and crew to support the IMR campaign. Recent progress in marine autonomy, autonomous perception, and acoustic communication, however, unlocks new approaches to subsea IMR that could substantially reduce the overall cost of these activities and enable novel inspection objectives. This work describes perception and behaviors enabled by a novel multi-agent autonomous system comprising an autonomous underwater vehicle (AUV) and unmanned surface vessel (USV). We describe a unique AUV configuration that enables real-time processing of sensor data streams both on vehicle and at surface. Results on the metrological performance achieved in a field test campaign and the autonomous hydrocarbon leak behavior that these capabilities enable are presented.