Real-time sonar-camera fusion lets underwater vehicles see through murky water

MIT and WHOI researchers introduced Sonar-MASt3R, a real-time opti-acoustic fusion approach that combines sonar data with camera imagery to generate centimeter-scale 3D maps of murky underwater environments. The system uses a coarse sonar map to guide the vehicle close enough for high-detail optical imaging, enabling safe navigation and object inspection in low-visibility water. Tested in a tank with varying turbidity, Sonar-MASt3R outperformed prior fusion methods and could aid underwater exploration, construction, maintenance, and deep-sea recovery, with plans to test in natural waters.