Scalable seagrass mapping using deep learning (GRASSMAP)

Research interests

• Autonomous robotic platforms allow detailed observations to be made over large areas in the ocean. For these systems to be useful, it is necessary to develop advanced sensing capabilities and methods to allow the robots to safely navigate and accurately localize themselves in complex, GPS denied environments. Once observations have been made, it’s necessary to interpret the large volumes of data that are gathered in an efficient and scalable way. For more information on research activities, please visit the Ocean Perception research website.
• Seafloor 3D visual reconstruction: Development of deep-sea imaging hardware and processing pipelines for calibration, localisation and 3D mapping of the seafloor with full-field uncertainty characterisation.
• BioCam (NERC NE/P020887/1): Development of a deep-sea, high-altitude seafloor imaging system for monitoring seafloor ecological variables as part of the Oceanids Marine Sensor Capital program. This project is a collaboration with Sonardyne International Ltd, the National Oceanography Centre and the ACFR University of SydneyAT-SEA (NERC NE/T010592/1): 3D visual survey of decommissioned seafloor infrastructure using a shore launched Autonomous Underwater Vehicle (Boaty McBoaface) as part of the INSITE program. This project is a collaboration with the National Oceanography Centre. Automated interpretation of data: Development of AI methods for rapid scalable interpretation of seafloor imagery.