Researchers from the Massachusetts Institute of Technology and the Woods Hole Oceanographic Institution have successfully developed Sonar-MASt3R, a novel underwater mapping system capable of generating real-time three-dimensional models in highly turbid environments. Led by MIT aerospace graduate student Amy Phung and WHOI senior scientist Richard Camilli, the project directly addresses a persistent bottleneck in autonomous marine operations: the rapid loss of optical visibility when underwater vehicles disturb seabed sediment. The system overcomes traditional limitations by fusing acoustic and optical data. While underwater cameras provide rich visual detail, they fail in low-light or high-turbidity conditions. Conversely, sonar penetrates murky water reliably to deliver precise distance and structural information, but lacks fine texture and object classification capabilities. Sonar-MASt3R integrates these modalities through a two-stage operational framework. Initially, a mounted sonar sensor conducts a lateral sweep to construct a coarse, absolute-scale topographical map. This acoustic baseline safely guides the vehicle toward target zones. Upon reaching optimal proximity, high-resolution optical images are captured and processed through an upgraded MASt3R algorithm, originally developed by Naver Labs Europe. The enhanced neural network rapidly computes dense local features and pixel-level relative depth, generating detailed point clouds without relying on external camera pose data. To maintain real-time performance, the system employs a dynamic keyframe selection strategy that filters redundant visual input, significantly reducing computational overhead. During controlled trials at the WHOI laboratory, the platform was subjected to eight varying levels of artificially induced turbidity. Results demonstrated consistent centimeter-level mapping accuracy and robust object recognition, even when optical sensors were completely obscured by suspended sediment. Laboratory conditions, characterized by confined-space acoustic reverberation, present a more stringent testing environment than open-water conditions, suggesting the system will perform effectively in actual maritime operations. The technology holds immediate utility for high-risk marine tasks, including unexploded ordnance clearance, deep-sea salvage, and critical infrastructure inspection, where operational pauses due to visibility loss can lead to mission delays or equipment damage. While the research has been featured at the IEEE International Conference on Robotics and Automation, the team is now transitioning to open-water validation to assess system resilience against dynamic currents, prolonged deployment cycles, and complex seabed topographies.