Final Report: NOAA Ocean Exploration NOFO NA21OAR0110196, Machine Learning for Automated Detection of Shipwreck Sites from Large Area Robotic Surveys

This project developed new technology and innovative methods to advance autonomous capabilities of marine robotic systems for search and survey of shipwreck sites. The main goal of this work was to increase efficiency and decrease cost for these missions. To achieve this goal, we proposed the following objectives: (i) to develop novel machine learning algorithms for online detection and ranking of potential targets of interest from sonar data collected through large area robotic surveys, (ii) to develop novel methods for efficient path planning of robotic surveys to collect high resolution imagery from potential sites of interest, and (iii) to validate proposed methods through large area robotic search and survey of shipwreck sites in the Thunder Bay National Marine Sanctuary (TBNMS) in Lake Huron.