Precise Underwater Localization Using Sonar and Electro-Optics (PULSE)

Underwater ship hull inspection platforms routinely check for damage, corrosion, and suspicious objects. Highly accurate position information is required to properly register hull inspection data with previously collected data to monitor the progression of structural anomalies and other changes and ensure complete hull coverage. Conventional underwater localization methods involving fixed-source signal triangulation are often ineffective at moored locations due to shallow water and interference from ship and dock structures. Dead-reckoning sensors onboard the inspection platform can provide local motion estimates, but small errors accumulate over time, eventually resulting in large position errors. Vision-based techniques, such as simultaneous localization and mapping (SLAM), have shown some promise for underwater platforms. However, the underwater environment presents several challenges to these techniques, such as poor visibility and limited scene structure. We propose a Precise Underwater Localization Using Sonar and Electro-Optics (PULSE) approach that uses optical and acoustic imaging sensors to offer greater flexibility under these conditions. The two sensors are also used together in a"stereo"configuration, enabling full 3D sensing to make the most of fine structural details on the hull. A particle-filter-based SLAM algorithm combines visual sensor data with motion data from dead-reckoning sensors to provide accurate location information in real-time.