Subsea Cable Search and Path Estimation Using Graph SLAM for AUV-Based Inspection

This research develops a method to efficiently search for and track subsea communication cables using an Autonomous Underwater Vehicle (AUVs). Inspection of subsea communication cables is a challenging task because they are narrow, making them difficult to detect, have significant route uncertainty that increases with depth, and can be buried for part of their length. The method is based on Simultaneous Localisation and Mapping (SLAM) using an initial map of the cable path that is assumed to be uncertain. Map-guided search patterns are first generated, where successful cable observations update the map using graph-based SLAM. We address the correspondence problem by defining a cable-relative coordinate frame that allows observations to be matched to map locations without the need for distinct cable features. Route uncertainty estimates are based on the location uncertainty of the AUV when observations are made, and physics-based catenary calculations for regions where the cable has not been observed. Updating cable route maps using this approach allows an AUV to efficiently recover a cable route even if it cannot be tracked for part of its length. Here we present the results of simulations that assume a camera-equipped AUV, and demonstrate robust performance of the method on five different cable routes, each with buried sections where the cable is undetectable and different initial cable route uncertainties.